research article delay-dependent finite-time filtering for...
TRANSCRIPT
Hindawi Publishing CorporationJournal of Applied MathematicsVolume 2013 Article ID 269091 13 pageshttpdxdoiorg1011552013269091
Research ArticleDelay-Dependent Finite-Time119867
infinFiltering for Markovian Jump
Systems with Different System Modes
Yong Zeng1 Jun Cheng1 Shouming Zhong2 and Xiucheng Dong3
1 School of Automation Engineering University of Electronic Science and Technology of China Chengdu Sichuan 611731 China2 School of Mathematical Sciences University of Electronic Science and Technology of China Chengdu Sichuan 611731 China3 Key Laboratory on Signal and Information Processing Xihua University Chengdu Sichuan 610039 China
Correspondence should be addressed to Jun Cheng jcheng6819126com
Received 4 March 2013 Accepted 16 April 2013
Academic Editor Qiankun Song
Copyright copy 2013 Yong Zeng et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited
This paper is concerned with the problem of delay-dependent finite-time119867infinfiltering for Markovian jump systems with different
system modes By using the new augmented multiple mode-dependent Lyapunov-Krasovskii functional and employing theproposed integrals inequalities in the derivation of our results a novel sufficient condition for finite-time boundness with an119867
infin
performance index is derived Particularly two different Markov processes have been considered for modeling the randomness ofsystem matrix and the state delay Based on the derived condition the 119867
infinfiltering problem is solved and an explicit expression
of the desired filter is also given the system trajectory stays within a prescribed bound during a specified time interval Finally anumerical example is given to illustrate the effectiveness and the potential of the proposed techniques
1 Introduction
Markovian jump systems were introduced by Krasovskiı andLidskiı [1] which can be described by a set of systemswith the transitions in a finite mode set In the past fewdecades there has been increasing interest in Markovianjump systems because this class of systems is appropriateto many physical systems which always go with randomfailures repairs and sudden environment disturbance [2ndash5]Such class of systems is a special class of stochastic hybridsystems with finite operation modes which may switch fromone to another at different time such as component failuressudden environmental disturbance and abrupt variationsof the operating points of a nonlinear system As a crucialfactor it is shown that such jumping can be determined by aMarkovian chain [6] For linear Markovian jumping systemsmany important issues have been studied extensively such asstability stabilization control synthesis and filter design [6ndash12] In finite operation modes Markovian jump systems are aspecial class of stochastic systems that can switch from one toanother at different time
It is worth pointing out that time delay is of interest tomany researchers because of the fact that time delay is often
encountered in various systems such as networked controlsystems chemical processes and communication systems Itis worth pointing out that time delay is one of the instabil-ity sources for dynamical systems and is a common phenom-enon inmany industrial and engineering systems Hence it isnot surprising that much effort has been made to investigateMarkovian jump systems with time delay during the last twodecades [13ndash15] The exponential stabilization of Markovianjump systems with time delay was firstly studied in [16]where the decay rate was estimated by solving linear matrixinequalities [17] However in the aforementioned works thenetwork-induced delays have been commonly assumed to bedeterministic which is fairly unrealistic since delays resultingfrom network transmissions are typically time varying [18ndash24]
Generally speaking the delay-dependent criterions areless conservative than delay-independent ones especiallywhen the time delay is small enough in Markovian jumpsystems Thus recent efforts were devoted to the delay-dependent Markovian jump systems stability analysis byemploying Lyapunov-Krasovskii functionals [25ndash33] How-ever in most thesis the time delay to be arbitrarily largeare allowed in criterion it always tends to be conservative
2 Journal of Applied Mathematics
Furthermore though the decay rate can be computed it is afixed value that one cannot adjust to deduce if a larger de-cay rate is possible Therefore how to obtain the improvedresults without increasing the computational burden hasgreatly improved the current study On the other hand thepractical problems which system described does not exceeda certain threshold over some finite time interval are con-sidered In finite-time interval finite-time stability is inves-tigated to address these transient performances of controlsystems Recently the concept of finite-time stability has beenrevisited in the light of linear matrix inequalities (LMIs) andLyapunov function theory and some results are obtained toensure that systems are finite-time stability or finite-timeboundness [34ndash50] To the best of our knowledge in mostof the works about Markovian jump systems with mode-dependent delay the delay mode is always assumed to be thesame as the system matrices mode However in real systemsthe delay mode may not be the same as that for jump inother system parameters In other words variations of delayusually depend on phenomena which may not cause abruptchanges in other systems parameters Therefore the work ofMarkovian jump systems with different system modes is notonly theoretically interesting and challenging but also veryimportant in practical applications
Motivated by the previous above discussions in this pa-per we present a new augmented Lyapunov functional fora class of Markovian jump systems with different systemmodes in order to reduce the possible conservativeness andcomputational burden some slack matrices are introduced[32] Several sufficient conditions are derived to guarantee thefinite-time stability and boundedness of the resulting closed-loop system We find that finite-time stability is an inde-pendent concept from Lyapunov stability and always can beaffected by switching behavior significantly and the finite-time boundness criteria can be tackled in the form of LMIsFinally a numerical example is presented to illustrate theeffectiveness of the developed techniques
Notations Throughout this paper we let 119875 gt 0 (119875 ge 0 119875 lt0 and 119875 le 0) denote a symmetric positive definite matrix 119875(positive semidefinite negative definite and negative semi-definite) For any symmetric matrix 119875 120582max(119875) and 120582min(119875)denote the maximum andminimum eigenvalues of matrix 119875respectivelyR119899 denotes the 119899-dimensional Euclidean spaceand R119899times119898 refers to the set of all 119899 times 119898 real matrices andN = 1 2 119873 The identity matrix of order 119899 is denotedas 119868
119899 lowast represents the elements below the main diagonal of a
symmetricmatrixThe superscripts ⊺ and minus1 stand formatrixtransposition and matrix inverse respectively
2 Preliminaries
In this paper we consider the followingMarkov jump systemdescribed by
(119905) = 119860119903119905
119909 (119905) + 119860120591119903119905
119909 (119905 minus 120591119904119905
(119905)) + 119863119903119905
120596 (119905)
119910 (119905) = 119862119910119903119905
119909 (119905) + 119862119910120591119903119905
119909 (119905 minus 120591119904119905
(119905)) + 119863119910119903119905
120596 (119905)
119911 (119905) = 119862119911119903119905
119909 (119905) + 119862119911120591119903119905
119909 (119905 minus 120591119904119905
(119905)) + 119863119911119903119905
120596 (119905)
119909 (119905) = 120593 (119905) 119905 = [minusℎ 0]
(1)
where 119909(119905) isinR119899 is the state vector of the system 119910(119905) isinR119897 isthemeasured output 119911(119905) isinR119902 is the controlled output120593(119905)119905 isin [minusℎ 0] are initial conditions of continuous state and 119903
0isin
N 1199040isin M = 1 2 119872 are initial conditions of mode
120596(119905) isin R119902 is the disturbance input satisfying the followingcondition
int
infin
0
120596⊺
(119905) 120596 (119905) 119889119905 le 119889 (2)
Let the random form processes 119903119905 119904
119905be the Markov sto-
chastic processes taking values on finite sets N = 1 2
119873 and M = 1 2 119872 with probability transition ratematrices Λ = 120582
119894119895 119894 119895 isin N and Π = 120587
119898119899 119898 119899 isin M The
transition probabilities from mode 119895 to mode 119895 for Markovprocess 119903
119905and from mode 119898 to mode 119899 for the Markov
process 119904119905in time ℎ are described as
Pr (119903119905+Δ
= 119895 | 119903119905= 119894) = 984858
119894119895+ 120582
119894119895Δ + 119900 (Δ)
Pr (119904119905+Δ
= 119899 | 119904119905= 119898) = 120577
119898119899+ 120587
119898119899Δ + 119900 (Δ)
(3)
where
984858119894119895=
0 if 119894 = 119895
1 if 119894 = 119895120577119898119899=
0 if 119898 = 119899
1 if 119898 = 119899
(4)
and Δ gt 0 120582119894119895ge 0 for 119894 = 119895 is the transition rate from mode
119894 at time 119905 to mode 119895 at time 119905 + Δ and
minus120582119894119894=
119873
sum
119895=1119895 = 119894
120582119894119895 (5)
for each mode 119894 isin N limΔrarr0
+
(119900(Δ)Δ) = 0 120587119898119899
ge 0 for119898 = 119899 is the transition rate from mode 119898 to mode 119899 at time119905 + Δ and
minus120587119898119898
=
119872
sum
119899=1119898 = 119899
120587119898119899 (6)
for each mode 119894 isin M limΔrarr0
+
(119900(Δ)Δ) = 0 For conve-nience we denote the Markov process 119903
119905and 119904
119905by 119894 and
119898 indices respectively 120591119898(119905) denotes the mode-dependent
time-varying state delay in the system and satisfies thefollowing condition
0 lt 120591119898(119905) le ℎ
119898lt infin
120591119898(119905) le 120583
119898 forall119898 isinM
(7)
where ℎ = maxℎ119894 119894 isin M is prescribed integer representing
the upper bounds of time-varying delay 120591119898(119905) Similarly 120583 =
max120583119898 119898 isin M is prescribed integer representing the up-
per bounds of time-varying delay 120591119898(119905) 119860
119903119905
119860120591119903119905
119863119903119905
119862119910119903119905
119862119910120591119903119905
119863119910119903119905
119862119911119903119905
119862119911120591119903119905
and 119863119911119903119905
are known mode-dependent
Journal of Applied Mathematics 3
matrices with appropriate dimensions functions of the ran-dom jumping process 119903
119905 and represent the nominal systems
for each 119903119905isin N For notation simplicity when the system
operates in the 119894-th mode (119903119905= 119894) 119860
119903119905
119860120591119903119905
119863119903119905
119862119910119903119905
119862119910120591119903119905
119863
119910119903119905
119862119911119903119905
119862119911120591119903119905
and119863119911119903119905
are denoted as119860119894119860
120591119894119863
119894119862
119910119894119862
119910120591119894
119863119910119894 119862
119911119894 119862
119911120591119894 and119863
119911119894 respectively
Here we are interested in designing a full-order filterdescribed by
119891(119905) = 119860
119891119903119905119904119905
119909119891(119905) + 119861
119891119903119905119904119905
119910 (119905)
119911119891(119905) = 119862
119891119903119905119904119905
119909119891(119905)
(8)
where 119909119891(119905) isin R119899 is the filter state 119911
119891(119896) isin R119902 and the
matrices 119860119891119903119905119904119905
119861119891119903119905119904119905
and 119862119891119903119905119904119905
are unknown filter param-eters to be designed
Augmenting the model of (1) to include the filter (8) weobtain the following filtering error system
120578 (119905) = 119860119903119905
120578 (119905) + 119861119903119905
120578 (119905 minus 120591119898(119905)) + 119863
119903119905
120596 (119905)
119890 (119905) = 119862119903119905
120578 (119905) + 119864119903119905
120578 (119905 minus 120591119898(119905))
(9)
where
120578 (119905) = [
119909 (119905)
119909119891(119905)] 119890 (119905) = 119911 (119905) minus 119911
119891(119905)
119860119903119905
= [
119860119903119905
0
119861119891119903119905119904119905
119862119910119903119905
119860119891119903119905119904119905
] 119861119903119905
= [
119860120591119903119905
119861119891119903119905119904119905
119862119910120591119903119905
]
119863119903119905
= [
119863119903119905
119861119891119903119905119904119905
119863119910119903119905
]
119862119903119905
= [119862119911119903119905
minus119862119891119903119905119904119905
] 119864119903119905
= [119862119911120591119903119905
0]
(10)
In order to more precisely describe the main objectivewe introduce the following definitions and Lemmas for theunderlying system
Definition 1 System (1) is said to be finite-time bounded withrespect to (119888
1 119888
2 119879 119877 119889) if condition (2) and the following
inequality hold
supminusℎle120592le0
E 120578⊺
(120592) 119877120578 (120592) 120578⊺
(120592) 119877 120578 (120592)
le 1198881997904rArr E 120578
⊺
(119905) 119877120578 (119905) lt 1198882
forall119905 isin [0 119879]
(11)
where 1198882gt 119888
1ge 0 and 119877 gt 0
Definition 2 Consider 119881(120578119905 119903
119905 119904
119905 119905 gt 0) as the stochastic
positive Lyapunov function its weak infinitesimal operatoris defined as
pound119881 (120578119905 119903
119905 119904
119905)
= limΔrarr0
1
Δ
[E 119881 (120578119905+Δ 119903
119905+Δ 119904
119905+Δ) | 120578
119905 119903
119905 119904
119905
minus119881 (120578119905 119903
119905 119904
119905)]
(12)
Definition 3 Given a constant 119879 gt 0 for all admissible120596(119905) subject to condition (2) under zero initial conditionsif the closed-loop Markovian jump system (1) is finite-timebounded and the control outputs satisfy condition (8) withattenuation 120574 gt 0
Eint119879
0
119890⊺
(119905) 119890 (119905) 119889119905 le 1205742
119890120578119879
Eint119879
0
120596⊺
(119905) 120596 (119905) 119889119905 (13)
Then the controller system (1) finite-time bounded withdisturbance attenuation 120574
Remark 4 It should be pointed out that the assumption ofzero initial condition in system (1) is only for the purposeof technical simplification in the derivation and it does notcause loss of generality In fact if this assumption is lost thesame control result can be obtained along the same lineexcept for adding extra manipulations in the derivation andextra terms in the control presentation However in real-world applications the initial condition of the underlyingsystem is generally not zero
Lemma 5 (see [32]) Let 119891119894 R119898
rarr R (119894 = 1 2 119873)
have positive values in an open subset D of R119898 Then the re-ciprocally convex combination of 119891
119894overD satisfies
min120573119894|120573119894gt0sum119894120573119894=1
sum
119894
1
120573119894
119891119894(119905) = sum
119894
119891119894(119905) +max
119892119894119895(119905)
sum
119894 = 119895
119892119894119895(119905)
119904119906119887119895119890119888119905 119905119900 119892119894119895R
119898
997888rarrR 119892119895119894(119905) = 119892
119894119895(119905)
[
119891119894(119905) 119892
119894119895(119905)
119892119894119895(119905) 119891
119895(119905)] ge 0
(14)
Lemma 6 (Schur Complement [17]) Given constant matrices119883119884119885 where119883 = 119883
⊺ and 0 lt 119884 = 119884⊺ then119883+119885⊺
119884minus1
119885 lt 0
if and only if
[
119883 119885⊺
lowast minus119884] lt 0 119900119903 [
minus119884 119885
lowast 119883] lt 0 (15)
3 Finite-Time 119867infin
Performance Analysis
Theorem 7 System (9) is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879) if there exist matrices 119875
119894119898gt 0 119876
119897119894gt 0 (119897 = 1
2) 119876 gt 0 119883119894119898gt 0 119883 gt 0 119884
119894119898gt 0 119884 gt 0119867 gt 0 119878
119894119898 scalars
1198881lt 119888
2 119879 gt 0120582
119904gt 0 (119904 = 1 2 10) 120578 gt 0 and Λ gt 0 such
that for all 119894 119895 isin N and 119898 119899 isin M the following inequalitieshold
sum
119895isinN119895 = 119894
120582119894119895(119876
1119895+ 119876
2119895) + sum
119899isinM119898 = 119899
120587119898119899119876
2119894minus 119876 lt 0 (16)
sum
119899isinM119898 = 119899
120587119898119899119883
119894119899+ sum
119895isinN119894 = 119895
120582119894119895119883
119895119898minus 119883 lt 0 (17)
4 Journal of Applied Mathematics
sum
119899isinM119898 = 119899
120587119898119899119884119894119899+ sum
119895isinN119894 = 119895
120582119894119895119884119895119898minus 119884 lt 0 (18)
[
[
[
119884119894119898
ℎ
119878119894119898
lowast
119884119894119898
ℎ
]
]
]
gt 0 (19)
Ξ1119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119872⊺
119894119898119875119894119898119863119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119872⊺
1198941198980
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119872⊺
119894119898119881119894119898119863119894
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119872119894119898119863119894
lowast lowast lowast lowast lowast minus120575119867
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(20)
Ξ2119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119872⊺
119894119898119875119894119898119863119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119873⊺
1198941198980
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119873⊺
119894119898119881119894119898119863119894
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119873119894119898119863119894
lowast lowast lowast lowast lowast minus120575119867
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(21)
1198881Λ + 119889120575120582
10
1
120578
(1 minus 119890minus120578119879
) lt 1205821119890minus120578119879
1198882 (22)
where
Ξ11119894119898
= sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898
+ 120575119875119894119898+ 119875
119894119898119860
119894+ 119860
⊺
119894119875119894119898+ 119890
120575ℎ
1198761119894
+ 119890120575ℎ
1198762119894+ ℎ119876
119890120575ℎ
minus 1
120575
119883119894119898
+
119890120575ℎ
minus 120575ℎ119890120575ℎ
minus 1
1205752
119883 +
119884119894119898
ℎ
Ξ12119894119898
= 119875119894119898119861119894minus
119884119894119898
ℎ
+ 119878119894119898
Ξ22119894119898
= 119903 (120583119898) 119876
2119894+
2119884119894119898
ℎ
minus 119878119894119898minus 119878
⊺
119894119898
Ξ44119894119898
= minus119881119894119898minus 119881
⊺
119894119898+
119890120575ℎ
minus 1
120575
119884119894119898+
119890120575ℎ
minus 120575ℎ119890120575ℎ
minus 1
1205752
119884
119903 (120591119898) =
minus (1 minus 120583119898) 119890
120575ℎ119894 if 120583
119898gt 1
minus (1 minus 120583119898) if 120583
119898le 1
Λ = 1205822+ ℎ119890
120575ℎ
(1205823+ 120582
4) + ℎ
2
119890120575ℎ
(1205825+ 120582
6+ 120582
8)
+
1
2
ℎ3
119890120575ℎ
(1205827+ 120582
9)
1205821= min
119894isinN119898isinM120582min (119894119898) 120582
2= max
119894isinN119898isinM120582max (119894119898)
1205823= max
119894isinN120582max (1198761119894
)
1205824= max
119894isinN120582max (1198762119894
) 1205825= 120582max (119876)
1205826= max
119894isinN119898isinM120582max (119883119894119898
) 1205827= 120582max (119883)
1205828= max
119894isinN119898isinM120582max (119894119898) 120582
9= 120582max ()
12058210= 120582max (119867)
119894119898= 119877
minus(12)
119875119894119898119877minus(12)
119876119897119894= 119877
minus(12)
119876119897119894119877minus(12)
(119897 = 1 2)
119876 = 119877minus(12)
119876119877minus(12)
119883119894119898= 119877
minus(12)
119883119894119898119877minus(12)
119883 = 119877minus(12)
119883119877minus(12)
119894119898= 119877
minus(12)
119884119894119898119877minus(12)
= 119877minus(12)
119884119877minus(12)
(23)
Proof First in order to cast our model into the framework oftheMarkov processes we define a new process (120578
119905 119903
119905 119904
119905) 119905 ge
0 by
120578119905(119904) = 120578 (119905 + 119904) 119904 isin [minusℎ 0] (24)
Now we consider the following Lyapunov-Krasovskiifunctional
119881 (120578119905 119903
119905 119904
119905) =
4
sum
119897=1
119881119897(120578
119905 119903
119905 119905) (25)
where
1198811(120578
119905 119903
119905 119904
119905) = 120578(119905)
⊺
119890120575119905
119875119903119905119904119905
120578 (119905)
1198812(120578
119905 119903
119905 119904
119905) = int
119905
119905minusℎ
119890120575(119904+ℎ)
120578⊺
(119904) 1198761119903119905
120578 (119904) 119889119904
+ int
119905
119905minus120591119904119905(119905)
119890120575(119904+ℎ)
120578⊺
(119904) 1198762119903119905
120578 (119904) 119889119904
+ int
0
minusℎ
int
119905
119905+120579
119890120575(119904+ℎ)
120578⊺
(119904) 119876120578 (119904) 119889119904 119889120579
1198813(120578
119905 119903
119905 119904
119905) = int
0
minusℎ
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904)119883119903119905119904119905
120578 (119904) 119889119904 119889120579
+ int
0
minusℎ
int
0
120579
int
119905
119905+120592
119890120575(119904minus120579)
120578⊺
(119904)119883120578 (119904) 119889119904 119889120592 119889120579
1198814(120578
119905 119903
119905 119904
119905) = int
0
minusℎ
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904) 119884119903119905119904119905
120578 (119904) 119889119904 119889120579
+ int
0
minusℎ
int
0
120579
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904) 119884 120578 (119904) 119889119904 119889120592 119889120579
(26)
Journal of Applied Mathematics 5
Then for each 119903119905= 119894 119904
119905= 119898 we have
pound1198811(120578
119905 119894 119898)
= limΔrarr0
+
1
Δ
E 120578⊺
(119905 + Δ) 119890120575(119905+Δ)
119875119903119905+Δ
119904119905+Δ
120578 (119905 + Δ)
minus120578⊺
(119905) 119890120575119905
119875119894119898120578 (119905)
= limΔrarr0
+
1
Δ
E
120578⊺
(119905 + Δ)[
[
(1 + 120582119894119894Δ + 119900 (Δ))
times (1 + 120587119898119898Δ + 119900 (Δ)) 119890
120575(119905+Δ)
119875119894119898
+ (1 + 120582119894119894Δ + 119900 (Δ))
times ( sum
119899isinM
(120587119898119899Δ + 119900 (Δ))) 119890
120575(119905+Δ)
119875119894119899
+ (1 + 120587119898119898Δ + 119900 (Δ))
times (sum
119895isinN
(120582119894119895Δ + 119900 (Δ))) 119890
120575(119905+Δ)
119875119895119898
+ ( sum
119899isinM
(120587119898119899Δ + 119900 (Δ)))
times(sum
119895isinN
(120582119894119895Δ + 119900 (Δ))) 119890
120575(119905+Δ)
119875119895119899
]
]
times 120578 (119905 + Δ)
minus 120578⊺
(119905) 119890120575119905
119875119894119898120578 (119905)
= 120575119890120575119905
120578⊺
(119905) 119875119894119898120578 (119905) + 2119890
120575119905
120578⊺
(119905) 119875119894119898120578 (119905)
+ 119890120575119905
120578⊺
(119905)( sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898)120578 (119905)
= 119890120575119905
120578⊺
(119905)( sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898+ 120575119875
119894119898
+ 119875119894119898119860
119894+ 119860
⊺
119894119875119894119898)120578 (119905)
+ 2119890120575119905
120578⊺
(119905) 119875119894119898119861119894120578 (119905 minus 120591
119898(119905)) + 2119890
120575119905
120578⊺
(119905) 119875119894119898119863
119894120596 (119905)
(27)pound119881
2(120578
119905 119894 119898)
= limΔrarr0
+
1
Δ
Eint119905+Δ
119905+Δminusℎ
120578⊺
(119904) 119890120575(119904+ℎ)
1198761119903119905+Δ
120578 (119904) 119889119904
minusint
119905
119905minusℎ
120578⊺
(119904) 119890120575(119904+ℎ)
1198761119894120578 (119904) 119889119904
+ limΔrarr0
+
1
Δ
Eint119905+Δ
119905+Δminus120591119904119905+Δ
(119905)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119903119905+Δ
120578 (119904) 119889119904
minusint
119905
119905minus120591119898(119905)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119894120578 (119904) 119889119904
+ limΔrarr0
+
1
Δ
Eint0
minusℎ
int
119905+Δ
119905+Δ+120579
119890120575(119904+ℎ)
120578⊺
(119904) 119876120578 (119904) 119889119904 119889120579
minusint
0
minusℎ
int
119905
119905+120579
119890120575(119904+ℎ)
120578⊺
(119904) 119876120578 (119904) 119889119904 119889120579
= limΔrarr0
+
1
Δ
Eint119905+Δ
119905+Δminusℎ
120578⊺
(119904) 119890120575(119904+ℎ)
1198761119894120578 (119904) 119889119904
+ int
119905+Δ
119905+Δminusℎ
120578⊺
(119904)
times sum
119895isinN
(120582119894119895Δ + 119900 (Δ))
times 119890120575(119904+ℎ)
1198761119894120578 (119904) 119889119904
minusint
119905
119905minusℎ
120578⊺
(119904) 1198761119894120578 (119904) 119889119904
+ limΔrarr0
+
1
Δ
E[int119905+Δ
119905+Δminus120591119895(119905+Δ)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119894120578 (119904) 119889119904
+ sum
119899isinM
(120587119898119899Δ + 119900 (Δ))
times int
119905
119905+Δminus120591119899(119905+Δ)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119894120578 (119904) 119889119904
+ int
119905+Δ
minus120591119895(119905 + Δ)
119905+Δ
120578⊺
(119904)
times sum
119895isinN
(120582119894119895Δ + 119900 (Δ))
times 119890120575(119904+ℎ)
1198762119895120578 (119904) 119889119904
minusint
119905
119905minus120591119898(119905)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119894120578 (119904) 119889119904]
+ 119890120575119905
120578⊺
(119905) ℎ119876120578 (119905) minus int
119905minusℎ
119890120575(119904+ℎ)
120578⊺
(119904) 119876120578 (119904) 119889119904
le 119890120575119905
120578⊺
(119905) (119890120575ℎ
1198761119894+ 119890
120575ℎ
1198762119894+ ℎ119876) 120578 (119905)
minus 119890120575119905
120578⊺
(119905 minus ℎ)1198761119894120578 (119905 minus ℎ)
minus (1 minus 120591119898(119905)) 119890
120575119905
120578⊺
(119905 minus 120591119898(119905)) 119876
2119894120578 (119905 minus 120591
119898(119905))
+ int
119905
119905minusℎ
119890120590(119904+ℎ)
120578⊺
(119904)( sum
119895isinN119895 = 119894
120582119894119895(119876
1119895+ 119876
2119895)
6 Journal of Applied Mathematics
+ sum
119899isinM119898 = 119899
120587119898119899119876
2119894minus 119876)
times 120578 (119904) 119889119904
(28)
Since
0 le 120591119898(119905) le ℎ
119898 (29)
we define
119903 (120591119898) =
minus (1 minus 120583119898) 119890
120575ℎ119898 if 120583
119898gt 1
minus (1 minus 120583119898) if 120583
119898le 1
(30)
Then
pound1198812(120578
119905 119894 119898) le 119890
120575119905
120578⊺
(119905) (119890120575ℎ
1198761119894+ 119890
120575ℎ
1198762119894+ ℎ119876) 120578 (119905)
minus 119890120575119905
120578⊺
(119905 minus ℎ)1198761119894120578 (119905 minus ℎ)
+ 119890120575119905
120578⊺
(119905 minus 120591119898(119905)) 119903 (120591
119898) 119876
2119894120578 (119905 minus 120591
119898(119905))
+ int
119905
119905minusℎ
119890120590(119904+ℎ)
120578⊺
(119904)
times ( sum
119895isinN119895 = 119894
120582119894119895(119876
1119895+ 119876
2119895)
+ sum
119899isinM119898 = 119899
120587119898119899119876
2119894minus 119876)
times 120578 (119904) 119889119904
(31)
Similar to the previous process we can obtain
pound1198813(120578
119905 119894 119898) le int
0
minusℎ
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904)
times ( sum
119899isinM119898 = 119899
120587119898119899119883
119894119899
+ sum
119895isinN119894 = 119895
120582119894119895119883
119895119898minus 119883)
times 120578 (119904) 119889119904 119889120579
+ 119890120575119905
120578⊺
(119905) 119883119894119898120578 (119905) int
0
minusℎ
119890minus120575120592
119889120592
minus 119890120575119905
int
119905
119905minusℎ
120578⊺
(119904)119883119894119898120578 (119904) 119889119904
+ 119890120575119905
120578⊺
(119905) 119883120578 (119905) int
0
minusℎ
int
0
120592
119890minus120575120592
119889120579119889120592
(32)
By using Lemma 5 it yields that
minusint
119905
119905minusℎ
120578⊺
(119904)119883119894119898120578 (119904) 119889119904 = minusint
119905
119905minus120591119898(119905)
120578⊺
(119904)119883119894119898120578 (119904) 119889119904
minus int
119905minus120591119898(119905)
119905minusℎ
120578⊺
(119904)119883119894119898120578 (119904) 119889119904
le minus120591119898(119905) 119880
⊺
1119883
1198941198981198801
minus (ℎ minus 120591119898(119905)) 119880
⊺
2119883
1198941198981198802
(33)
where
1198801=
1
120591119898(119905)
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904
1198802=
1
ℎ minus 120591119898(119905)
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904
lim120591119898(119905)rarr0
1
120591119898(119905)
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904 = 120578 (119905)
lim120591119898(119905)rarrℎ
1
ℎ minus 120591119898(119905)
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904 = 120578 (119905 minus ℎ)
(34)
From the Newton-Leibniz formula the following equa-tion is true for any matrices119872
119894119898 119873
119894119898 and 119881
119894119898with appro-
priate dimensions
(2120591119898(119905) 119880
⊺
1119872
119894119898+ 2 (ℎ minus 120591
119898(119905)) 119880
⊺
2119873
119894119898+ 2 120578
⊺
(119905) 119881119894119898)
times [minus 120578 (119905) + 119860119894120578 (119905) + 119861
120591119894120578 (119905 minus 120591
119898(119905)) + 119863
119894120596 (119905)] = 0
(35)
pound1198814(120578
119905 119894 119898)
le int
0
minusℎ
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904)
times ( sum
119899isinM119898 = 119899
120587119898119899119884119894119899
+ sum
119895isinN119894 = 119895
120582119894119895119884119895119898minus 119884)
times 120578 (119904) 119889119904 119889120579
+ 119890120575119905
120578⊺
(119905) 119884119894119898120578 (119905) int
0
minusℎ
119890minus120575120592
119889120592
minus 119890120575119905
int
119905
119905minusℎ
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
+ 119890120575119905
120578⊺
(119905) 119884 120578 (119905) int
0
minusℎ
int
0
120592
119890minus120575120592
119889120579 119889120592
(36)
Journal of Applied Mathematics 7
From Lemma 5 it yields that
minus int
119905
119905minusℎ
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
= minusint
119905
119905minus120591119898(119905)
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
minus int
119905minus120591119898(119905)
119905minusℎ
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
le minus
ℎ
120591119898(119905)
[int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904]
⊺
119884119894119898
ℎ
[int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904]
minus
ℎ
ℎ minus 120591119898(119905)
[int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904]
⊺
119884119894119898
ℎ
[int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904]
le minus
[
[
[
[
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904
]
]
]
]
⊺
[
[
[
119884119894119898
ℎ
119878119894119898
lowast
119884119894119898
ℎ
]
]
]
[
[
[
[
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904
]
]
]
]
(37)
From (25)ndash(37) we can eventually obtain
pound119881 (120578119905 119903
119905 119904
119905) minus 120575120596
⊺
(119905)119867120596 (119905) le 119890120575119905
120585⊺
(119905) Ξ119894119898120585 (119905) (38)
where
120585⊺
(119905) = [120578⊺
(119905) 120578⊺
(119905 minus 120591119898(119905)) 120578
⊺
(119905 minus ℎ) 120578⊺
(119905) 119880⊺
1 119880
⊺
2 120596
⊺
(119905)]
Ξ119894119898=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860⊺
119894119881
⊺
119894119898120591119898(119905) 119860
⊺
119894119872
⊺
119894119898(ℎ minus 120591
119898(119905)) 119860
⊺
119894119873
⊺
119894119898119875119894119898119863
119894
lowast Ξ22119894119898
119878119894119898minus
119884119894119898
ℎ
119860⊺
120591119894119881
⊺
119894119898120591119898(119905) 119860
⊺
120591119894119872
⊺
119894119898(ℎ minus 120591
119898(119905)) 119860
⊺
120591119894119873
⊺
1198941198980
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minus120591119898(119905)119872
⊺
119894119898minus (ℎ minus 120591
119898(119905))119873
⊺
119894119898119881119894119898119863
119894
lowast lowast lowast lowast minus120591119898(119905) 119883
1198941198980 120591
119898(119905)119872
119894119898119863
119894
lowast lowast lowast lowast lowast minus (ℎ minus 120591119898(119905))119883
119894119898(ℎ minus 120591
119898(119905))119873
119894119898119863
119894
lowast lowast lowast lowast lowast lowast minus120575119867
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
(39)
The LMIs (20) and (21) lead to 120591119898(119905) rarr ℎ and 120591
119898(119905) rarr
0 respectively It is easy to see that Ξ1119894119898
results from Ξ119894119898|
120591119898(119905) = ℎ and Ξ
119894119898|120591119898(119905) = 0 Thus we obtain
E pound119881 (120578119905 119903
119905 119904
119905) le 120578E [119881 (120578
119905 119903
119905 119904
119905)] + 120575120596
⊺
(119905)119867120596 (119905)
(40)
Multiplying the aforementioned inequality by 119890minus120578119905 we canget
E pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] le 119890
minus120578119905
120575120596⊺
(119905)119867120596 (119905) (41)
By integrating the aforementioned inequality between 0and 119905 it follows that
119890minus120578119905
E [119881 (120578119905 119903
119905 119904
119905)] minus E [119881 (120578
0 119903
0 119904
0)]
le 120575int
119905
0
119890minus120578119904
120596⊺
(119904)119867120596 (119904) 119889119904
(42)
Denote that 119894119898
= 119877minus(12)
119875119894119898119877minus(12) 119876
119904119894=
119877minus(12)
119876119904119894119877minus(12)
(119904 = 1 2) 119876 = 119877minus(12)
119876119877minus(12)
119883119894119898
= 119877minus(12)
119883119894119898119877minus(12) 119883 = 119877
minus(12)
119883119877minus(12)
119894119898
= 119877minus(12)
119884119894119898119877minus(12) and = 119877
minus(12)
119884119877minus(12) it yields
that
E [119881 (1205780 119903
0 119904
0)]
le max119894isinN119898isinM
120582max (119894119898) 120578⊺
(0) 119877120578 (0)
+ (max119894isinN
120582max (1198761119894) +max
119894isinN120582max (1198762119894
)) 119890120575ℎ
times int
0
minusℎ
119890120575119904
120578⊺
(119904) 119877120578 (119904) 119889119904
+ 119890120575ℎ
120582max (119876)int0
minusℎ
int
0
120579
119890120575119904
120578⊺
(119904) 119877120578 (119904) 119889119904
+ 119890120575ℎ max
119894isinN119898isinM120582max (119883119894119898
)
times int
0
minusℎ
int
0
120579
119890minus120575120579
120578⊺
(119904) 119877120578 (119904) 119889119904 119889120579
+ 119890120575ℎ
120582max (119883)
times int
0
minusℎ
int
0
120579
int
0
120592
119890minus120575120579
120578⊺
(119904) 119877120578 (119904) 119889119904 119889120579 119889120592
+ 119890120575ℎ max
119894isinN119898isinM120582max (119894119898)
8 Journal of Applied Mathematics
times int
0
minusℎ
int
0
120579
119890minus120575120579
120578⊺
(119904) 119877 120578 (119904) 119889119904 119889120579
+ 119890120575ℎ
120582max ()
times int
0
minusℎ
int
0
120579
int
0
120592
119890minus120575120579
120578⊺
(119904) 119877 120578 (119904) 119889119904 119889120579 119889120592
le max119894isinN119898isinM
120582max (119894119898)
+ ℎ119890120575ℎ
(max119894isinN
120582max (1198761119894+
119894isinNmax120582max (1198762119894
)))
+ ℎ2
119890120575ℎ
( max119894isinN119898isinM
120582max (119883119894119898) + 120582max (119876)
+ max119894isinN119898isinM
120582max (119894119898))
+
1
2
ℎ3
119890120575ℎ
(120582max (119883) + 120582max ())
times supminusℎle119904le0
120578⊺
(119904) 119877120578 (119904) 120578⊺
(119904) 119877 120578 (119904)
= 1198881Λ
(43)
For given 120578 gt 0 and 0 le 119905 le 119879 we have
E [119881 (120578119905 119903
119905 119904
119905)] le E [119890
120578119905
119881 (1205780 119903
0 119904
0)]
+ 119890120578119905
120575int
119905
0
119890minus120578119904
120596⊺
(119904)119867120596 (119904) 119889119904
le 119890120578119879
1198881Λ + 119889120575119890
120578119879
120582max (119867)int119879
0
119890minus120578119904
119889119904
le 119890120578119879
1198881Λ + 119889120575120582
10
1
120578
(1 minus 119890minus120578119879
)
(44)
On the other hand it follows from (25) that
E [119881 (120578119905 119903
119905 119904
119905)] ge E [120578
⊺
(119905) 119890120582119905
119875119894119898120578 (119905)]
ge max119894isinN
120582min (119875119894119898)E [120578⊺
(119905) 119877120578 (119905)]
= 1205821E [120578
⊺
(119905) 119877120578 (119905)]
(45)
It can be derived from (43)ndash(45) that
E [120578⊺
(119905) 119877120578 (119905)] le
119890120578119879
1205821
1198881Λ + 119889120575120582
10
1
120578
(1 minus 119890minus120578119879
) (46)
From (22) and (46) we have
E [120578⊺
(119905) 119877120578 (119905)] lt 1198882 (47)
Then the system is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879)
Remark 8 In this paper 120591119898(119905) and 120591
119898(119905) may have different
upper bounds in various delay intervals satisfying (7) respec-tively However in previous work such as [20 21] 120591
119898(119905) and
120591119898(119905) are enlarged to 120591
119898(119905) le ℎ = maxℎ
119894 119898 isin M and
120591119898(119905) le 120583 = max120583
119898 119898 isin M respectively which may lead
to conservativeness inevitably However the previouse casecan be taken fully into account by employing the Lyapunov-Krasovskii functional (25)
Remark 9 When dealing with termminusint
119905
119905minusℎ
120578⊺
(119904)119884119894119898120578(119904)119889119904 the
convex combination is not employed Lemma 5 is used in thispaper and then the free-weighting matrices-dependent nulladd items are necessary to be introduced in our proof whichlead to the decrease of the number of LMIs and LMIs scalardecision variables
Remark 10 The feature of this paper is the way to dealwith the integral term Many researchers have enlarged thederivative of the Lyapunov functional in order to deal with theintegral term in mathematical operations In this paper wepropose a novel delay-dependent sufficient criterion whichensures that the Markovian jump system with different modesystems is finite-time stable
Remark 11 It should be pointed out that the novelty of theLyapunov functional (25) lies in distinct Lyapunov matrices(119875
119894119898 119883
119894119898 119884
119894119898) which is chosen for different system modes
119894 (119894 = 1 2 119873) and119898 (119898 = 1 2 119872)
Theorem 12 System (9) is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879) if there exist matrices 119875
119894gt 0 119876
119897119894gt 0 (119897 = 1 2)
119876 gt 0 119883119894119898gt 0 119883 gt 0 119884
119894119898gt 0 119884 gt 0 119878
119894119898 scalars 119888
1lt 119888
2
119879 gt 0 120582119904gt 0 (119904 = 1 2 9) 120578 gt 0 120574 gt 0 and Λ gt 0 such
that for all 119894 119895 isinN and119898 119899 isinM and (16)ndash(19) the followinginequalities hold
Σ1119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119872⊺
119894119898119875119894119898119863119894
119862
⊺
119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119872⊺
1198941198980 119864
⊺
119894
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119872⊺
119894119898119881119894119898119863119894
0
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119872119894119898119863119894
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(48)
Σ2119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119873⊺
119894119898119875119894119898119863119894
119862
⊺
119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119873⊺
1198941198980 119864
⊺
119894
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119873⊺
119894119898119881119894119898119863119894
0
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119873119894119898119863119894
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(49)
1198881Λ + 119889120575120574
21
120578
(1 minus 119890minus120578119879
) lt 1205821119890minus120578119879
1198882 (50)
Journal of Applied Mathematics 9
Proof We now consider the 119867infin
performance of system (9)Select the same Lyapunov-Krasovskii functional as Theo-rem 7 it yields that
pound119881 (120578119905 119903
119905 119904
119905) + 119890
⊺
(119905) 119890 (119905) minus 1205742
120596⊺
(119905) 120596 (119905) le 120585⊺
(119905) Σ119897119894119898120585 (119905)
(119897 = 1 2)
(51)
It follows from (49)-(50) that
E pound119881 (120578119905 119903
119905 119904
119905)
le E [120578119881 (120578119905 119903
119905 119904
119905)] + 120574
2
120596⊺
(119905) 120596 (119905) minus E [119890⊺
(119905) 119890 (119905)]
(52)
Multiplying the aforementioned inequality by 119890minus120578119905 onehas
E pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] le 119890
minus120578119905
[1205742
120596⊺
(119905) 120596 (119905) minus 119890⊺
(119905) 119890 (119905)]
(53)
In zero initial condition and E[119881(120578119905 119903
119905 119904
119905)] gt 0 by inte-
grating the aforementioned inequality between 0 and 119879 wecan get
int
119879
0
119890minus120578119905
[1205742
120596⊺
(119905) 120596 (119905) minus 119890⊺
(119905) 119890 (119905)] 119889119905
le Eint119879
0
pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] 119889119905 le 119881 (120578
0 119903
0 119904
0) = 0
(54)
Using Dynkins formula it results that
E [int119879
0
119890minus120578120592
119890⊺
(120592) 119890 (120592) 119889120592] le 1205742
E [int119879
0
119890minus120578120592
120596⊺
(120592) 120596 (120592) 119889120592]
(55)
Then it yields that
E [int119879
0
119890⊺
(120592) 119890 (120592) 119889120592] le 1205742
119890120578119879
E [int119879
0
120596⊺
(120592) 120596 (120592) 119889120592] (56)
Thus it is concluded by Definition 3 that system (9) is fi-nite-time bounded with an 119867
infinperformance 120574 This com-
pletes the proof
Remark 13 From the proof process of Theorems 7 and 12it is easy to see that neither bounding technique for cross-terms nor model transformation is involved In other wordsthe obtained result is expected to be less conservative
Remark 14 The Lyapunov asymptotic stability and finite-time stability of a class of system are independent conceptsA Lyapunov asymptotic stability system may not be finite-time stability Moreover finite-time stability system may alsonot be Lyapunov asymptotic stabilityThere exist some resultson Lyapunov stability while finite-time stability also needsour full investigation which was neglected by most previouswork
4 Finite-Time 119867infin
Filtering
Theorem 15 System (9) is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879) if there exist matrices 119875
119894119898gt 0119872
119894119898 119873
119894119898 119881
119894119898
119876119897119894gt 0 (119897 = 1 2) 119876 gt 0 119883
119894119898gt 0 119883 gt 0 119884
119894119898gt 0 119884 gt 0
119860119891119894119898
119861119891119894119898
119862119891119894119898
119872119860119894119898
119872119861119894119898
119873119860119894119898
119873119861119894119898
119881119860119894119898
119881119861119894119898
119878119894119898
scalars 1198881lt 119888
2 119879 gt 0 120590
119904gt 0 (119904 = 1 2 9) 120575 gt 0 120578 gt 0
and Λ gt 0 such that for all 119894 119895 isin N and 119898 119899 isin M thefollowing inequalities hold
119875119894119898= [
1198751119894119898
1198752119894119898
1198752119894119898
1198752119894119898
] gt 0 119872119894119898= [
1198721119894119898
1198722119894119898
1198722119894119898
1198722119894119898
] (57)
119873119894119898= [
1198731119894119898
1198732119894119898
1198732119894119898
1198732119894119898
] 119881119894119898= [
1198811119894119898
1198812119894119898
1198812119894119898
1198812119894119898
] (58)
Γ1119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Γ11119894119898
Γ12119894119898
minus119878119894119898
Γ14119894119898
Γ15119894119898
Γ16119894119898
Γ17119894119898
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
Γ24119894119898
Γ25119894119898
0 [119862⊺
119911120591119894
0]
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119872⊺
119894119898Γ46119894119898
0
lowast lowast lowast lowast minusℎ119883119894119898
Γ56119894119898
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(59)
Γ2119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Γ11119894119898
Γ12119894119898
minus119878119894119898
Γ14119894119898
Γ1015840
15119894119898Γ16119894119898
Γ17119894119898
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
Γ24119894119898
Γ1015840
251198941198980 [
119862⊺
119911120591119894
0]
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119873⊺
119894119898Γ1015840
461198941198980
lowast lowast lowast lowast minusℎ119883119894119898
Γ1015840
561198941198980
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(60)where
Γ11119894119898
= sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898+ 119890
120575ℎ
1198761119894+ 119890
120575ℎ
1198762119894+ ℎ119876 +
119890120575ℎ
minus 1
120575
119883119894119898+
119890120575ℎ
minus 120575ℎ119890120575ℎ
minus 1
1205752
119883
+
119884119894119898
ℎ
+ [
1205751198751119894119898
+ 1198751119894119898119860
119894+ 119860
⊺
1198941198751119894119898
+ 119861119891119894119898119862119910119894+ 119862
⊺
119910119894119861
⊺
119891119894119898120575119875
2119894119898+ 119860
119891119894119898+ 119860
⊺
119891119894119898
1205751198752119894119898
+ 1198752119894119898119860
119894+ 119860
⊺
1198941198752119894119898
+ 119861119891119894119898119862119910119894+ 119862
⊺
119910119894119861
⊺
119891119894119898120575119875
2119894119898+ 119860
119891119894119898+ 119860
⊺
119891119894119898
]
(61)
10 Journal of Applied Mathematics
Γ12119894119898
= minus
119884119894119898
ℎ
+ 119878119894119898+ [
1198751119894119898119860
120591119894+ 119861
119891119894119898119862119910120591119894
1198752119894119898119860
120591119894+ 119861
119891119894119898119862119910120591119894
]
Γ14119894119898
= [
119860⊺
119894119881
⊺
1119894119898+ 119862
⊺
119910119894119881
⊺
119861119894119898119881
⊺
119860119894119898
119860⊺
119894119881
⊺
2119894119898+ 119862
⊺
119910119894119881
⊺
119861119894119898119881
⊺
119860119894119898
]
Γ15119894119898
= [
ℎ119860⊺
119894119872
⊺
1119894119898+ ℎ119862
⊺
119910119894119872
⊺
119861119894119898ℎ119872
⊺
119860119894119898
ℎ119860⊺
119894119872
⊺
2119894119898+ ℎ119862
⊺
119910119894119872
⊺
119861119894119898ℎ119872
⊺
119860119894119898
]
Γ1015840
15119894119898= [
ℎ119860⊺
119894119873
⊺
1119894119898+ ℎ119862
⊺
119910119894119873
⊺
119861119894119898ℎ119873
⊺
119860119894119898
ℎ119860⊺
119894119873
⊺
2119894119898+ ℎ119862
⊺
119910119894119873
⊺
119861119894119898ℎ119873
⊺
119860119894119898
]
Γ16119894119898
= [
1198751119894119898119863
119894+ 119861
119891119894119898119863
119910119894
1198752119894119898119863
119894+ 119861
119891119894119898119863
119910119894
]
Γ17119894119898
= [
119862⊺
119911119894
minus119862119891119894119898
]
Γ24119894119898
= [
119860⊺
120591119894119881
⊺
1119894119898+ 119862
⊺
119910120591119894119881
⊺
119861119894119898
119860⊺
120591119894119881
⊺
2119894119898+ 119862
⊺
119910120591119894119881
⊺
119861119894119898
]
Γ25119894119898
= [
ℎ119860⊺
120591119894119872
⊺
1119894119898+ ℎ119862
⊺
119910120591119894119872
⊺
119861119894119898
ℎ119860⊺
120591119894119872
⊺
2119894119898+ ℎ119862
⊺
119910120591119894119872
⊺
119861119894119898
]
Γ1015840
25119894119898= [
ℎ119860⊺
120591119894119873
⊺
1119894119898+ ℎ119862
⊺
119910120591119894119873
⊺
119861119894119898
ℎ119860⊺
120591119894119873
⊺
2119894119898+ ℎ119862
⊺
119910120591119894119873
⊺
119861119894119898
]
Γ46119894119898
= [
1198811119894119898119863
119894+ 119881
119861119894119898119863
119910119894
1198812119894119898119863
119894+ 119881
119861119894119898119863
119910119894
]
Γ56119894119898
= [
ℎ1198721119894119898119863
119894+ ℎ119872
119861119894119898119863
119910119894
ℎ1198722119894119898119863
119894+ ℎ119872
119861119894119898119863
119910119894
]
Γ1015840
56119894119898= [
ℎ1198731119894119898119863
119894+ ℎ119873
119861119894119898119863
119910119894
ℎ1198732119894119898119863
119894+ ℎ119873
119861119894119898119863
119910119894
]
(62)
Then a desired filter can be chosen with parameters as
119860119891119894119898
= 119875minus1
2119894119898119860
119891119894119898 119861
119891119894119898= 119875
minus1
2119894119898119861119891119894119898
119862119891119894119898
= 119862119891119894119898
(63)
Proof We denote that
119875119894119898= [
1198751119894119898
1198752119894119898
1198752119894119898
1198752119894119898
] (64)
The term 119875119894119898119860
119894can be rewritten as
119875119894119898119860
119894= [
1198751119894119898119860
119894+ 119875
21198941198981198611198911198941198981198621199101198941198752119894119898119860
119891119894119898
1198752119894119898119860
119894+ 119875
21198941198981198611198911198941198981198621199101198941198752119894119898119860
119891119894119898
] (65)
Similarly we have
119875119894119898119861119894= [
1198751119894119898119860
120591119894+ 119875
2119894119898119861119891119894119898119862119910120591119894
1198752119894119898119860
120591119894+ 119875
2119894119898119861119891119894119898119862119910120591119894
]
119875119894119898119863
119894= [
1198751119894119898119863
119894+ 119875
2119894119898119861119891119894119898119863
119910119894
1198752119894119898119863
119894+ 119875
2119894119898119861119891119894119898119863
119910119894
]
(66)
Define 119860119891119894119898
= 1198752119894119898119860
119891119894119898 119861
119891119894119898= 119875
2119894119898119861119891119894119898
119862119891119894119898
= 119862119891119894119898
119872
119860119894119898= 119872
2119894119898119860
119891119894119898119872
119861119894119898= 119872
2119894119898119861119891119894119898
119873119860119894119898
= 1198732119894119898119860
119891119894119898
119873119861119894119898
= 1198732119894119898119861119891119894119898
119881119860119894119898
= 1198812119894119898119860
119891119894119898 and 119881
119861119894119898= 119881
2119894119898119861119891119894119898
Therefore if (59) and (60) hold system (9) is finite-timebounded with a prescribed 119867
infinperformance index 120574 The
proof is completed
Remark 16 In many actual applications the minimum valueof 1205742min is of interest In Theorem 12 with a fixed 120582 120574min canbe obtained through the following optimization procedure
min 1205742
st (48)ndash(50) (67)
In Theorem 15 as for finite-time stability and bounded-ness once the state bound 119888
2is not ascertained theminimum
value 1198882min is of interest With a fixed 120582 define 120582
1= 1 then
the following optimization problem can be formulated to getminimum value 119888
2min
min 1205891205742
+ (1 minus 120589) 1198882
st (59)-(60) and (50)
(68)
where 120589 is weighted factor and 120589 isin [0 1]
5 Illustrative Example
Example 17 Consider that the Markovian jump system andthe delay mode switching are governed by a Markov processwith the following transition rates
Λ = [
minus07 07
09 minus09] Π = [
minus1 1
12 minus12] (69)
as well as with the following parameters
1198601= [
minus35 086
minus064 minus325] 119860
2= [
minus25 034
14 minus002]
1198601205911= [
minus08 minus13
minus07 minus22] 119860
1205912= [
minus28 05
minus08 minus14]
1198631= 119863
2= [
01
02] 119862
1199101= [04 07]
1198621199102= [05 08] 119862
1199101205911= [01 02]
1198621199101205912
= [01 01]
1198631199101= 119863
1199102= 01 119862
1199111= [01 005]
1198621199112= [02 009] 119862
1199111205911= [03 06]
1198621199111205912
= [04 08] 1198631199111= 119863
1199112= 005
(70)
Journal of Applied Mathematics 11
Then we choose 119877 = 119868 119879 = 2 1198881= 1 and 119889 = 001
throughTheorem 15 it yields that themode-dependent filtersare as follows
11986011989111
= [
minus61254 15565
minus02347 minus03763]
11986011989112
= [
minus62682 14338
minus03552 minus05327]
11986011989121
= [
minus89214 24223
minus15476 minus06234]
11986011989122
= [
minus84638 28237
minus13545 minus04322]
11986111989111
= [
51465 minus28516
minus91216 95171]
11986111989112
= [
54203 minus23121
minus93156 96332]
11986111989121
= [
46193 minus11984
minus162397 163006]
11986111989122
= [
46512 minus10311
minus161132 162312]
11986211989111
= [00294 minus00397]
11986211989112
= [00271 minus003368]
11986211989121
= [01163 minus01432]
11986211989122
= [01342 minus03672]
(71)
This paper deals with the finite-time filter design problemfor a class of Markovian jump systems particularly two dif-ferentMarkov processes are considered formodeling the ran-domness of system matrix and the state delay Then throughthe numerical example we can see that results in this paperare feasible which further verified the correctness of our the-ory Therefore the paper shorten this gap
6 Conclusions
In this paper we have examined the problems of finite-time119867
infinfiltering for a class of Markovian jump systems with dif-
ferent system modes Based on a novel approach a sufficientcondition is derived such that the closed-loop Markovianjump system is finite-time bounded and satisfies a prescribedlevel of119867
infindisturbance attenuation in a finite time interval
Finally a numerical example is also given to illustrate theeffectiveness of the proposed design approach It should benoted that one of future research topics would be to investi-gate the problems of fault detection and fault tolerant controlfor time-varying Markovian jump systems with incompleteinformation over a finite-time horizon
Acknowledgments
The authors would like to thank the associate editor and theanonymous reviewers for their detailed comments and sug-gestions This work was supported by the Fund of SichuanProvincial Key Laboratory of Signal and Information Pro-cessing Xihua University (SZJJ2009-002 and SGXZD0101-10-1) and National Basic Research Program of China(2010CB732501)
References
[1] N N Krasovskiı and E A Lidskiı ldquoAnalytical design ofcontrollers in systems with random attributesmdashI Statementof the problem method of solvingrdquo Automation and RemoteControl vol 22 pp 1021ndash1025 1961
[2] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[3] L Wu X Su and P Shi ldquoSliding mode control with bounded1198712gain performance of Markovian jump singular time-delay
systemsrdquo Automatica vol 48 no 8 pp 1929ndash1933 2012[4] H Gao Z Fei J Lam and B Du ldquoFurther results on expo-
nential estimates of Markovian jump systems with mode-de-pendent time-varying delaysrdquo IEEE Transactions on AutomaticControl vol 56 no 1 pp 223ndash229 2011
[5] Z Wu H Su and J Chu ldquo119867infinfiltering for singular Markovian
jump systems with time delayrdquo International Journal of Robustand Nonlinear Control vol 20 no 8 pp 939ndash957 2010
[6] P Tino M Cernansky and L Benuskova ldquoMarkovian architec-tural bias of recurrent neural networksrdquo IEEE Transactions onNeural Networks vol 15 no 1 pp 6ndash15 2004
[7] H Dong Z Wang D W C Ho and H Gao ldquoRobust 119867infin
filtering for Markovian jump systems with randomly occurringnonlinearities and sensor saturation the finite-horizon caserdquoIEEE Transactions on Signal Processing vol 59 no 7 pp 3048ndash3057 2011
[8] Q Ma S Xu and Y Zou ldquoStability and synchronization forMarkovian jump neural networks with partly unknown tran-sition probabilitiesrdquo Neurocomputing vol 74 pp 3404ndash33412011
[9] H Zhao S Xu and Y Zou ldquoRobust119867infinfiltering for uncertain
Markovian jump systems with mode-dependent distributeddelaysrdquo International Journal of Adaptive Control and SignalProcessing vol 24 no 1 pp 83ndash94 2010
[10] Y Zhang S Xu and B Zhang ldquoDiscrete-time fuzzy Markovianjump systems with time-varying delaysrdquo IEEE Transactions onFuzzy Systems vol 17 no 2 pp 411ndash420 2009
[11] Z Wu H Su and J Chu ldquoState estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 no 10ndash12 pp 2247ndash2254 2010
[12] Q Zhu and J Cao ldquoRobust exponential stability of markovianjump impulsive stochastic Cohen-Grossberg neural networkswithmixed time delaysrdquo IEEE Transactions onNeural Networksvol 21 no 8 pp 1314ndash1325 2010
[13] H Huang G Feng and X Chen ldquoStability and stabilizationof Markovian jump systems with time delay via new Lyapunovfunctionalsrdquo IEEE Transactions on Circuits and Systems I vol59 no 10 pp 2413ndash2421 2012
12 Journal of Applied Mathematics
[14] Q Zhu and J Cao ldquoStability of Markovian jump neural net-works with impulse control and time varying delaysrdquoNonlinearAnalysis Real World Applications vol 13 no 5 pp 2259ndash22702012
[15] M Sun J Lam S Xu andY Zou ldquoRobust exponential stabiliza-tion for Markovian jump systems with mode-dependent inputdelayrdquo Automatica vol 43 no 10 pp 1799ndash1807 2007
[16] M Mohammadian A Abolmasoumi and H Momeni ldquo119867infin
mode-independent filter design for Markovian jump geneticregulatory networks with time-varying delaysrdquo Neurocomput-ing vol 87 pp 10ndash18 2012
[17] S Boyd L El Ghaoui E Feron and V Balakrishnan LinearMatrix Inequalities in SystemandControlTheory vol 15 of SIAMStudies in Applied Mathematics SIAM Philadelphia Pa USA1994
[18] S Mou H GaoW Qiang and K Chen ldquoNew delay-dependentexponential stability for neural networks with time delayrdquo IEEETransactions on Systems Man and Cybernetics Part B vol 38no 2 pp 571ndash576 2008
[19] O M Kwon J Park S Lee and E Cha ldquoAnalysis on delay-dependent stability for neural networks with time-varyingdelaysrdquo Neurocomputing vol 103 pp 114ndash120 2013
[20] L Hu H Gao andW X Zheng ldquoNovel stability of cellular neu-ral networkswith interval time-varying delayrdquoNeuralNetworksvol 21 no 10 pp 1458ndash1463 2008
[21] H Zhang Z Liu G B Huang and ZWang ldquoNovel weighting-delay-based stability criteria for recurrent neural networks withtime-varying delayrdquo IEEE Transactions on Neural Networks vol21 no 1 pp 91ndash106 2010
[22] Z Zuo C Yang and YWang ldquoA newmethod for stability anal-ysis of recurrent neural networks with interval time-varyingdelayrdquo IEEE Transactions on Neural Networks vol 21 no 2 pp339ndash344 2010
[23] P Park JWKo andC Jeong ldquoReciprocally convex approach tostability of systems with time-varying delaysrdquo Automatica vol47 no 1 pp 235ndash238 2011
[24] D Zhang L Yu and W Zhang ldquoDelay-dependent fault detec-tion for switched linear systems with time-varying delays-theaverage dwell time approachrdquo Signal Processing vol 91 pp 832ndash840 2011
[25] Z Fei H Gao and P Shi ldquoNew results on stabilization ofMark-ovian jump systems with time delayrdquoAutomatica vol 45 no 10pp 2300ndash2306 2009
[26] H R Karimi ldquoRobust delay-dependent 119867infin
control of uncer-tain time-delay systems with mixed neutral discrete anddistributed time-delays and Markovian switching parametersrdquoIEEE Transactions on Circuits and Systems I vol 58 no 8 pp1910ndash1923 2011
[27] M Luo G Liu and S Zhong ldquoRobust fault detection ofMarkovian jump systems with different systemmodesrdquoAppliedMathematical Modelling vol 37 no 7 pp 5001ndash5012 2013
[28] B Zhang andY Li ldquoExponential1198712minus119871
infinfiltering for distributed
delay systems with Markovian jumping parametersrdquo SignalProcessing vol 93 no 1 pp 206ndash216 2013
[29] A Abolmasoumi and H Momeni ldquoRobust observer-based119867infin
control of a Markovian jump system with different delay andsystem modesrdquo International Journal of Control Automationand Systems vol 9 no 4 pp 768ndash776 2011
[30] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[31] S Ma and E K Boukas ldquoRobust 119867infin
filtering for uncertaindiscrete Markov jump singular systems with mode-dependenttime delayrdquo IET Control Theory amp Applications vol 3 no 3 pp351ndash361 2009
[32] Y ChenW Bi andW Li ldquoStability analysis for neural networkswith time-varying delay a more general delay decompositionapproachrdquo Neurocomputing vol 73 no 4ndash6 pp 853ndash857 2010
[33] Z Shu J Lam and S Xu ldquoRobust stabilization of Markoviandelay systems with delay-dependent exponential estimatesrdquoAutomatica vol 42 no 11 pp 2001ndash2008 2006
[34] X Huang W Lin and B Yang ldquoGlobal finite-time stabilizationof a class of uncertain nonlinear systemsrdquo Automatica vol 41no 5 pp 881ndash888 2005
[35] C Qian and J Li ldquoGlobal finite-time stabilization by outputfeedback for planar systems without observable linearizationrdquoIEEE Transactions on Automatic Control vol 50 no 6 pp 885ndash890 2005
[36] F Amato R Ambrosino C Cosentino and G De TommasildquoInput-output finite time stabilization of linear systemsrdquo Auto-matica vol 46 no 9 pp 1558ndash1562 2010
[37] Y Hong ldquoFinite-time stabilization and stabilizability of a classof controllable systemsrdquo Systems amp Control Letters vol 46 no4 pp 231ndash236 2002
[38] F Amato and M Ariola ldquoFinite-time control of discrete-timelinear systemsrdquo IEEETransactions onAutomatic Control vol 50no 5 pp 724ndash729 2005
[39] Z Zuo Y Liu Y Wang and H Li ldquoFinite-time stochasticstability and stabilisation of linear discrete-time Markovianjump systems with partly unknown transition probabilitiesrdquoIET Control Theory amp Applications vol 6 no 10 pp 1522ndash15262012
[40] X Luan F Liu and P Shi ldquoFinite-time filtering for non-linearstochastic systems with partially known transition jump ratesrdquoIET Control Theory amp Applications vol 4 no 5 pp 735ndash7452010
[41] F Amato M Ariola and C Cosentino ldquoFinite-time control ofdiscrete-time linear systems analysis and design conditionsrdquoAutomatica vol 46 no 5 pp 919ndash924 2010
[42] X Lin H Du and S Li ldquoFinite-time boundedness and 1198712-
gain analysis for switched delay systems with norm-boundeddisturbancerdquo Applied Mathematics and Computation vol 217no 12 pp 5982ndash5993 2011
[43] J Lin S Fei and Z Gao ldquoStabilization of discrete-time switchedsingular time-delay systems under asynchronous switchingrdquoJournal of the Franklin Institute vol 349 no 5 pp 1808ndash18272012
[44] H Song L Yu D Zhang and W-A Zhang ldquoFinite-time 119867infin
control for a class of discrete-time switched time-delay systemswith quantized feedbackrdquoCommunications in Nonlinear Scienceand Numerical Simulation vol 17 no 12 pp 4802ndash4814 2012
[45] Z Zuo H Li Y Liu and Y Wang ldquoOn finite-time stochasticstability and stabilization ofMarkovian jump systems subject topartial information on transition probabilitiesrdquoCircuits Systemsand Signal Processing 2012
[46] H Liu Y Shen and X Zhao ldquoDelay-dependent observer-based119867
infinfinite-time control for switched systems with time-varying
delayrdquoNonlinear Analysis Hybrid Systems vol 6 no 3 pp 885ndash898 2012
[47] Y Zhang C Liu and X Mu ldquoRobust finite-time119867infincontrol of
singular stochastic systems via static output feedbackrdquo AppliedMathematics and Computation vol 218 no 9 pp 5629ndash56402012
Journal of Applied Mathematics 13
[48] J Cheng H Zhu S Zhong Y Zhang and Y Zeng ldquoFinite-timeStabilization of 119867
infinfiltering for switched stochastic systemsrdquo
Circuits Systems and Signal Processing 2012[49] J Cheng H Zhu S Zhong and Y Zhang ldquoFinite-time bound-
ness of 119867infin
filtering for switching discrete-time systemsrdquo In-ternational Journal of Control Automation and Systems vol 10pp 1129ndash1135 2012
[50] Z Xiang C Qiao and M S Mahmoud ldquoFinite-time analysisand119867
infincontrol for switched stochastic systemsrdquo Journal of the
Franklin Institute vol 349 no 3 pp 915ndash927 2012
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
2 Journal of Applied Mathematics
Furthermore though the decay rate can be computed it is afixed value that one cannot adjust to deduce if a larger de-cay rate is possible Therefore how to obtain the improvedresults without increasing the computational burden hasgreatly improved the current study On the other hand thepractical problems which system described does not exceeda certain threshold over some finite time interval are con-sidered In finite-time interval finite-time stability is inves-tigated to address these transient performances of controlsystems Recently the concept of finite-time stability has beenrevisited in the light of linear matrix inequalities (LMIs) andLyapunov function theory and some results are obtained toensure that systems are finite-time stability or finite-timeboundness [34ndash50] To the best of our knowledge in mostof the works about Markovian jump systems with mode-dependent delay the delay mode is always assumed to be thesame as the system matrices mode However in real systemsthe delay mode may not be the same as that for jump inother system parameters In other words variations of delayusually depend on phenomena which may not cause abruptchanges in other systems parameters Therefore the work ofMarkovian jump systems with different system modes is notonly theoretically interesting and challenging but also veryimportant in practical applications
Motivated by the previous above discussions in this pa-per we present a new augmented Lyapunov functional fora class of Markovian jump systems with different systemmodes in order to reduce the possible conservativeness andcomputational burden some slack matrices are introduced[32] Several sufficient conditions are derived to guarantee thefinite-time stability and boundedness of the resulting closed-loop system We find that finite-time stability is an inde-pendent concept from Lyapunov stability and always can beaffected by switching behavior significantly and the finite-time boundness criteria can be tackled in the form of LMIsFinally a numerical example is presented to illustrate theeffectiveness of the developed techniques
Notations Throughout this paper we let 119875 gt 0 (119875 ge 0 119875 lt0 and 119875 le 0) denote a symmetric positive definite matrix 119875(positive semidefinite negative definite and negative semi-definite) For any symmetric matrix 119875 120582max(119875) and 120582min(119875)denote the maximum andminimum eigenvalues of matrix 119875respectivelyR119899 denotes the 119899-dimensional Euclidean spaceand R119899times119898 refers to the set of all 119899 times 119898 real matrices andN = 1 2 119873 The identity matrix of order 119899 is denotedas 119868
119899 lowast represents the elements below the main diagonal of a
symmetricmatrixThe superscripts ⊺ and minus1 stand formatrixtransposition and matrix inverse respectively
2 Preliminaries
In this paper we consider the followingMarkov jump systemdescribed by
(119905) = 119860119903119905
119909 (119905) + 119860120591119903119905
119909 (119905 minus 120591119904119905
(119905)) + 119863119903119905
120596 (119905)
119910 (119905) = 119862119910119903119905
119909 (119905) + 119862119910120591119903119905
119909 (119905 minus 120591119904119905
(119905)) + 119863119910119903119905
120596 (119905)
119911 (119905) = 119862119911119903119905
119909 (119905) + 119862119911120591119903119905
119909 (119905 minus 120591119904119905
(119905)) + 119863119911119903119905
120596 (119905)
119909 (119905) = 120593 (119905) 119905 = [minusℎ 0]
(1)
where 119909(119905) isinR119899 is the state vector of the system 119910(119905) isinR119897 isthemeasured output 119911(119905) isinR119902 is the controlled output120593(119905)119905 isin [minusℎ 0] are initial conditions of continuous state and 119903
0isin
N 1199040isin M = 1 2 119872 are initial conditions of mode
120596(119905) isin R119902 is the disturbance input satisfying the followingcondition
int
infin
0
120596⊺
(119905) 120596 (119905) 119889119905 le 119889 (2)
Let the random form processes 119903119905 119904
119905be the Markov sto-
chastic processes taking values on finite sets N = 1 2
119873 and M = 1 2 119872 with probability transition ratematrices Λ = 120582
119894119895 119894 119895 isin N and Π = 120587
119898119899 119898 119899 isin M The
transition probabilities from mode 119895 to mode 119895 for Markovprocess 119903
119905and from mode 119898 to mode 119899 for the Markov
process 119904119905in time ℎ are described as
Pr (119903119905+Δ
= 119895 | 119903119905= 119894) = 984858
119894119895+ 120582
119894119895Δ + 119900 (Δ)
Pr (119904119905+Δ
= 119899 | 119904119905= 119898) = 120577
119898119899+ 120587
119898119899Δ + 119900 (Δ)
(3)
where
984858119894119895=
0 if 119894 = 119895
1 if 119894 = 119895120577119898119899=
0 if 119898 = 119899
1 if 119898 = 119899
(4)
and Δ gt 0 120582119894119895ge 0 for 119894 = 119895 is the transition rate from mode
119894 at time 119905 to mode 119895 at time 119905 + Δ and
minus120582119894119894=
119873
sum
119895=1119895 = 119894
120582119894119895 (5)
for each mode 119894 isin N limΔrarr0
+
(119900(Δ)Δ) = 0 120587119898119899
ge 0 for119898 = 119899 is the transition rate from mode 119898 to mode 119899 at time119905 + Δ and
minus120587119898119898
=
119872
sum
119899=1119898 = 119899
120587119898119899 (6)
for each mode 119894 isin M limΔrarr0
+
(119900(Δ)Δ) = 0 For conve-nience we denote the Markov process 119903
119905and 119904
119905by 119894 and
119898 indices respectively 120591119898(119905) denotes the mode-dependent
time-varying state delay in the system and satisfies thefollowing condition
0 lt 120591119898(119905) le ℎ
119898lt infin
120591119898(119905) le 120583
119898 forall119898 isinM
(7)
where ℎ = maxℎ119894 119894 isin M is prescribed integer representing
the upper bounds of time-varying delay 120591119898(119905) Similarly 120583 =
max120583119898 119898 isin M is prescribed integer representing the up-
per bounds of time-varying delay 120591119898(119905) 119860
119903119905
119860120591119903119905
119863119903119905
119862119910119903119905
119862119910120591119903119905
119863119910119903119905
119862119911119903119905
119862119911120591119903119905
and 119863119911119903119905
are known mode-dependent
Journal of Applied Mathematics 3
matrices with appropriate dimensions functions of the ran-dom jumping process 119903
119905 and represent the nominal systems
for each 119903119905isin N For notation simplicity when the system
operates in the 119894-th mode (119903119905= 119894) 119860
119903119905
119860120591119903119905
119863119903119905
119862119910119903119905
119862119910120591119903119905
119863
119910119903119905
119862119911119903119905
119862119911120591119903119905
and119863119911119903119905
are denoted as119860119894119860
120591119894119863
119894119862
119910119894119862
119910120591119894
119863119910119894 119862
119911119894 119862
119911120591119894 and119863
119911119894 respectively
Here we are interested in designing a full-order filterdescribed by
119891(119905) = 119860
119891119903119905119904119905
119909119891(119905) + 119861
119891119903119905119904119905
119910 (119905)
119911119891(119905) = 119862
119891119903119905119904119905
119909119891(119905)
(8)
where 119909119891(119905) isin R119899 is the filter state 119911
119891(119896) isin R119902 and the
matrices 119860119891119903119905119904119905
119861119891119903119905119904119905
and 119862119891119903119905119904119905
are unknown filter param-eters to be designed
Augmenting the model of (1) to include the filter (8) weobtain the following filtering error system
120578 (119905) = 119860119903119905
120578 (119905) + 119861119903119905
120578 (119905 minus 120591119898(119905)) + 119863
119903119905
120596 (119905)
119890 (119905) = 119862119903119905
120578 (119905) + 119864119903119905
120578 (119905 minus 120591119898(119905))
(9)
where
120578 (119905) = [
119909 (119905)
119909119891(119905)] 119890 (119905) = 119911 (119905) minus 119911
119891(119905)
119860119903119905
= [
119860119903119905
0
119861119891119903119905119904119905
119862119910119903119905
119860119891119903119905119904119905
] 119861119903119905
= [
119860120591119903119905
119861119891119903119905119904119905
119862119910120591119903119905
]
119863119903119905
= [
119863119903119905
119861119891119903119905119904119905
119863119910119903119905
]
119862119903119905
= [119862119911119903119905
minus119862119891119903119905119904119905
] 119864119903119905
= [119862119911120591119903119905
0]
(10)
In order to more precisely describe the main objectivewe introduce the following definitions and Lemmas for theunderlying system
Definition 1 System (1) is said to be finite-time bounded withrespect to (119888
1 119888
2 119879 119877 119889) if condition (2) and the following
inequality hold
supminusℎle120592le0
E 120578⊺
(120592) 119877120578 (120592) 120578⊺
(120592) 119877 120578 (120592)
le 1198881997904rArr E 120578
⊺
(119905) 119877120578 (119905) lt 1198882
forall119905 isin [0 119879]
(11)
where 1198882gt 119888
1ge 0 and 119877 gt 0
Definition 2 Consider 119881(120578119905 119903
119905 119904
119905 119905 gt 0) as the stochastic
positive Lyapunov function its weak infinitesimal operatoris defined as
pound119881 (120578119905 119903
119905 119904
119905)
= limΔrarr0
1
Δ
[E 119881 (120578119905+Δ 119903
119905+Δ 119904
119905+Δ) | 120578
119905 119903
119905 119904
119905
minus119881 (120578119905 119903
119905 119904
119905)]
(12)
Definition 3 Given a constant 119879 gt 0 for all admissible120596(119905) subject to condition (2) under zero initial conditionsif the closed-loop Markovian jump system (1) is finite-timebounded and the control outputs satisfy condition (8) withattenuation 120574 gt 0
Eint119879
0
119890⊺
(119905) 119890 (119905) 119889119905 le 1205742
119890120578119879
Eint119879
0
120596⊺
(119905) 120596 (119905) 119889119905 (13)
Then the controller system (1) finite-time bounded withdisturbance attenuation 120574
Remark 4 It should be pointed out that the assumption ofzero initial condition in system (1) is only for the purposeof technical simplification in the derivation and it does notcause loss of generality In fact if this assumption is lost thesame control result can be obtained along the same lineexcept for adding extra manipulations in the derivation andextra terms in the control presentation However in real-world applications the initial condition of the underlyingsystem is generally not zero
Lemma 5 (see [32]) Let 119891119894 R119898
rarr R (119894 = 1 2 119873)
have positive values in an open subset D of R119898 Then the re-ciprocally convex combination of 119891
119894overD satisfies
min120573119894|120573119894gt0sum119894120573119894=1
sum
119894
1
120573119894
119891119894(119905) = sum
119894
119891119894(119905) +max
119892119894119895(119905)
sum
119894 = 119895
119892119894119895(119905)
119904119906119887119895119890119888119905 119905119900 119892119894119895R
119898
997888rarrR 119892119895119894(119905) = 119892
119894119895(119905)
[
119891119894(119905) 119892
119894119895(119905)
119892119894119895(119905) 119891
119895(119905)] ge 0
(14)
Lemma 6 (Schur Complement [17]) Given constant matrices119883119884119885 where119883 = 119883
⊺ and 0 lt 119884 = 119884⊺ then119883+119885⊺
119884minus1
119885 lt 0
if and only if
[
119883 119885⊺
lowast minus119884] lt 0 119900119903 [
minus119884 119885
lowast 119883] lt 0 (15)
3 Finite-Time 119867infin
Performance Analysis
Theorem 7 System (9) is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879) if there exist matrices 119875
119894119898gt 0 119876
119897119894gt 0 (119897 = 1
2) 119876 gt 0 119883119894119898gt 0 119883 gt 0 119884
119894119898gt 0 119884 gt 0119867 gt 0 119878
119894119898 scalars
1198881lt 119888
2 119879 gt 0120582
119904gt 0 (119904 = 1 2 10) 120578 gt 0 and Λ gt 0 such
that for all 119894 119895 isin N and 119898 119899 isin M the following inequalitieshold
sum
119895isinN119895 = 119894
120582119894119895(119876
1119895+ 119876
2119895) + sum
119899isinM119898 = 119899
120587119898119899119876
2119894minus 119876 lt 0 (16)
sum
119899isinM119898 = 119899
120587119898119899119883
119894119899+ sum
119895isinN119894 = 119895
120582119894119895119883
119895119898minus 119883 lt 0 (17)
4 Journal of Applied Mathematics
sum
119899isinM119898 = 119899
120587119898119899119884119894119899+ sum
119895isinN119894 = 119895
120582119894119895119884119895119898minus 119884 lt 0 (18)
[
[
[
119884119894119898
ℎ
119878119894119898
lowast
119884119894119898
ℎ
]
]
]
gt 0 (19)
Ξ1119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119872⊺
119894119898119875119894119898119863119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119872⊺
1198941198980
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119872⊺
119894119898119881119894119898119863119894
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119872119894119898119863119894
lowast lowast lowast lowast lowast minus120575119867
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(20)
Ξ2119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119872⊺
119894119898119875119894119898119863119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119873⊺
1198941198980
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119873⊺
119894119898119881119894119898119863119894
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119873119894119898119863119894
lowast lowast lowast lowast lowast minus120575119867
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(21)
1198881Λ + 119889120575120582
10
1
120578
(1 minus 119890minus120578119879
) lt 1205821119890minus120578119879
1198882 (22)
where
Ξ11119894119898
= sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898
+ 120575119875119894119898+ 119875
119894119898119860
119894+ 119860
⊺
119894119875119894119898+ 119890
120575ℎ
1198761119894
+ 119890120575ℎ
1198762119894+ ℎ119876
119890120575ℎ
minus 1
120575
119883119894119898
+
119890120575ℎ
minus 120575ℎ119890120575ℎ
minus 1
1205752
119883 +
119884119894119898
ℎ
Ξ12119894119898
= 119875119894119898119861119894minus
119884119894119898
ℎ
+ 119878119894119898
Ξ22119894119898
= 119903 (120583119898) 119876
2119894+
2119884119894119898
ℎ
minus 119878119894119898minus 119878
⊺
119894119898
Ξ44119894119898
= minus119881119894119898minus 119881
⊺
119894119898+
119890120575ℎ
minus 1
120575
119884119894119898+
119890120575ℎ
minus 120575ℎ119890120575ℎ
minus 1
1205752
119884
119903 (120591119898) =
minus (1 minus 120583119898) 119890
120575ℎ119894 if 120583
119898gt 1
minus (1 minus 120583119898) if 120583
119898le 1
Λ = 1205822+ ℎ119890
120575ℎ
(1205823+ 120582
4) + ℎ
2
119890120575ℎ
(1205825+ 120582
6+ 120582
8)
+
1
2
ℎ3
119890120575ℎ
(1205827+ 120582
9)
1205821= min
119894isinN119898isinM120582min (119894119898) 120582
2= max
119894isinN119898isinM120582max (119894119898)
1205823= max
119894isinN120582max (1198761119894
)
1205824= max
119894isinN120582max (1198762119894
) 1205825= 120582max (119876)
1205826= max
119894isinN119898isinM120582max (119883119894119898
) 1205827= 120582max (119883)
1205828= max
119894isinN119898isinM120582max (119894119898) 120582
9= 120582max ()
12058210= 120582max (119867)
119894119898= 119877
minus(12)
119875119894119898119877minus(12)
119876119897119894= 119877
minus(12)
119876119897119894119877minus(12)
(119897 = 1 2)
119876 = 119877minus(12)
119876119877minus(12)
119883119894119898= 119877
minus(12)
119883119894119898119877minus(12)
119883 = 119877minus(12)
119883119877minus(12)
119894119898= 119877
minus(12)
119884119894119898119877minus(12)
= 119877minus(12)
119884119877minus(12)
(23)
Proof First in order to cast our model into the framework oftheMarkov processes we define a new process (120578
119905 119903
119905 119904
119905) 119905 ge
0 by
120578119905(119904) = 120578 (119905 + 119904) 119904 isin [minusℎ 0] (24)
Now we consider the following Lyapunov-Krasovskiifunctional
119881 (120578119905 119903
119905 119904
119905) =
4
sum
119897=1
119881119897(120578
119905 119903
119905 119905) (25)
where
1198811(120578
119905 119903
119905 119904
119905) = 120578(119905)
⊺
119890120575119905
119875119903119905119904119905
120578 (119905)
1198812(120578
119905 119903
119905 119904
119905) = int
119905
119905minusℎ
119890120575(119904+ℎ)
120578⊺
(119904) 1198761119903119905
120578 (119904) 119889119904
+ int
119905
119905minus120591119904119905(119905)
119890120575(119904+ℎ)
120578⊺
(119904) 1198762119903119905
120578 (119904) 119889119904
+ int
0
minusℎ
int
119905
119905+120579
119890120575(119904+ℎ)
120578⊺
(119904) 119876120578 (119904) 119889119904 119889120579
1198813(120578
119905 119903
119905 119904
119905) = int
0
minusℎ
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904)119883119903119905119904119905
120578 (119904) 119889119904 119889120579
+ int
0
minusℎ
int
0
120579
int
119905
119905+120592
119890120575(119904minus120579)
120578⊺
(119904)119883120578 (119904) 119889119904 119889120592 119889120579
1198814(120578
119905 119903
119905 119904
119905) = int
0
minusℎ
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904) 119884119903119905119904119905
120578 (119904) 119889119904 119889120579
+ int
0
minusℎ
int
0
120579
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904) 119884 120578 (119904) 119889119904 119889120592 119889120579
(26)
Journal of Applied Mathematics 5
Then for each 119903119905= 119894 119904
119905= 119898 we have
pound1198811(120578
119905 119894 119898)
= limΔrarr0
+
1
Δ
E 120578⊺
(119905 + Δ) 119890120575(119905+Δ)
119875119903119905+Δ
119904119905+Δ
120578 (119905 + Δ)
minus120578⊺
(119905) 119890120575119905
119875119894119898120578 (119905)
= limΔrarr0
+
1
Δ
E
120578⊺
(119905 + Δ)[
[
(1 + 120582119894119894Δ + 119900 (Δ))
times (1 + 120587119898119898Δ + 119900 (Δ)) 119890
120575(119905+Δ)
119875119894119898
+ (1 + 120582119894119894Δ + 119900 (Δ))
times ( sum
119899isinM
(120587119898119899Δ + 119900 (Δ))) 119890
120575(119905+Δ)
119875119894119899
+ (1 + 120587119898119898Δ + 119900 (Δ))
times (sum
119895isinN
(120582119894119895Δ + 119900 (Δ))) 119890
120575(119905+Δ)
119875119895119898
+ ( sum
119899isinM
(120587119898119899Δ + 119900 (Δ)))
times(sum
119895isinN
(120582119894119895Δ + 119900 (Δ))) 119890
120575(119905+Δ)
119875119895119899
]
]
times 120578 (119905 + Δ)
minus 120578⊺
(119905) 119890120575119905
119875119894119898120578 (119905)
= 120575119890120575119905
120578⊺
(119905) 119875119894119898120578 (119905) + 2119890
120575119905
120578⊺
(119905) 119875119894119898120578 (119905)
+ 119890120575119905
120578⊺
(119905)( sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898)120578 (119905)
= 119890120575119905
120578⊺
(119905)( sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898+ 120575119875
119894119898
+ 119875119894119898119860
119894+ 119860
⊺
119894119875119894119898)120578 (119905)
+ 2119890120575119905
120578⊺
(119905) 119875119894119898119861119894120578 (119905 minus 120591
119898(119905)) + 2119890
120575119905
120578⊺
(119905) 119875119894119898119863
119894120596 (119905)
(27)pound119881
2(120578
119905 119894 119898)
= limΔrarr0
+
1
Δ
Eint119905+Δ
119905+Δminusℎ
120578⊺
(119904) 119890120575(119904+ℎ)
1198761119903119905+Δ
120578 (119904) 119889119904
minusint
119905
119905minusℎ
120578⊺
(119904) 119890120575(119904+ℎ)
1198761119894120578 (119904) 119889119904
+ limΔrarr0
+
1
Δ
Eint119905+Δ
119905+Δminus120591119904119905+Δ
(119905)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119903119905+Δ
120578 (119904) 119889119904
minusint
119905
119905minus120591119898(119905)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119894120578 (119904) 119889119904
+ limΔrarr0
+
1
Δ
Eint0
minusℎ
int
119905+Δ
119905+Δ+120579
119890120575(119904+ℎ)
120578⊺
(119904) 119876120578 (119904) 119889119904 119889120579
minusint
0
minusℎ
int
119905
119905+120579
119890120575(119904+ℎ)
120578⊺
(119904) 119876120578 (119904) 119889119904 119889120579
= limΔrarr0
+
1
Δ
Eint119905+Δ
119905+Δminusℎ
120578⊺
(119904) 119890120575(119904+ℎ)
1198761119894120578 (119904) 119889119904
+ int
119905+Δ
119905+Δminusℎ
120578⊺
(119904)
times sum
119895isinN
(120582119894119895Δ + 119900 (Δ))
times 119890120575(119904+ℎ)
1198761119894120578 (119904) 119889119904
minusint
119905
119905minusℎ
120578⊺
(119904) 1198761119894120578 (119904) 119889119904
+ limΔrarr0
+
1
Δ
E[int119905+Δ
119905+Δminus120591119895(119905+Δ)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119894120578 (119904) 119889119904
+ sum
119899isinM
(120587119898119899Δ + 119900 (Δ))
times int
119905
119905+Δminus120591119899(119905+Δ)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119894120578 (119904) 119889119904
+ int
119905+Δ
minus120591119895(119905 + Δ)
119905+Δ
120578⊺
(119904)
times sum
119895isinN
(120582119894119895Δ + 119900 (Δ))
times 119890120575(119904+ℎ)
1198762119895120578 (119904) 119889119904
minusint
119905
119905minus120591119898(119905)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119894120578 (119904) 119889119904]
+ 119890120575119905
120578⊺
(119905) ℎ119876120578 (119905) minus int
119905minusℎ
119890120575(119904+ℎ)
120578⊺
(119904) 119876120578 (119904) 119889119904
le 119890120575119905
120578⊺
(119905) (119890120575ℎ
1198761119894+ 119890
120575ℎ
1198762119894+ ℎ119876) 120578 (119905)
minus 119890120575119905
120578⊺
(119905 minus ℎ)1198761119894120578 (119905 minus ℎ)
minus (1 minus 120591119898(119905)) 119890
120575119905
120578⊺
(119905 minus 120591119898(119905)) 119876
2119894120578 (119905 minus 120591
119898(119905))
+ int
119905
119905minusℎ
119890120590(119904+ℎ)
120578⊺
(119904)( sum
119895isinN119895 = 119894
120582119894119895(119876
1119895+ 119876
2119895)
6 Journal of Applied Mathematics
+ sum
119899isinM119898 = 119899
120587119898119899119876
2119894minus 119876)
times 120578 (119904) 119889119904
(28)
Since
0 le 120591119898(119905) le ℎ
119898 (29)
we define
119903 (120591119898) =
minus (1 minus 120583119898) 119890
120575ℎ119898 if 120583
119898gt 1
minus (1 minus 120583119898) if 120583
119898le 1
(30)
Then
pound1198812(120578
119905 119894 119898) le 119890
120575119905
120578⊺
(119905) (119890120575ℎ
1198761119894+ 119890
120575ℎ
1198762119894+ ℎ119876) 120578 (119905)
minus 119890120575119905
120578⊺
(119905 minus ℎ)1198761119894120578 (119905 minus ℎ)
+ 119890120575119905
120578⊺
(119905 minus 120591119898(119905)) 119903 (120591
119898) 119876
2119894120578 (119905 minus 120591
119898(119905))
+ int
119905
119905minusℎ
119890120590(119904+ℎ)
120578⊺
(119904)
times ( sum
119895isinN119895 = 119894
120582119894119895(119876
1119895+ 119876
2119895)
+ sum
119899isinM119898 = 119899
120587119898119899119876
2119894minus 119876)
times 120578 (119904) 119889119904
(31)
Similar to the previous process we can obtain
pound1198813(120578
119905 119894 119898) le int
0
minusℎ
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904)
times ( sum
119899isinM119898 = 119899
120587119898119899119883
119894119899
+ sum
119895isinN119894 = 119895
120582119894119895119883
119895119898minus 119883)
times 120578 (119904) 119889119904 119889120579
+ 119890120575119905
120578⊺
(119905) 119883119894119898120578 (119905) int
0
minusℎ
119890minus120575120592
119889120592
minus 119890120575119905
int
119905
119905minusℎ
120578⊺
(119904)119883119894119898120578 (119904) 119889119904
+ 119890120575119905
120578⊺
(119905) 119883120578 (119905) int
0
minusℎ
int
0
120592
119890minus120575120592
119889120579119889120592
(32)
By using Lemma 5 it yields that
minusint
119905
119905minusℎ
120578⊺
(119904)119883119894119898120578 (119904) 119889119904 = minusint
119905
119905minus120591119898(119905)
120578⊺
(119904)119883119894119898120578 (119904) 119889119904
minus int
119905minus120591119898(119905)
119905minusℎ
120578⊺
(119904)119883119894119898120578 (119904) 119889119904
le minus120591119898(119905) 119880
⊺
1119883
1198941198981198801
minus (ℎ minus 120591119898(119905)) 119880
⊺
2119883
1198941198981198802
(33)
where
1198801=
1
120591119898(119905)
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904
1198802=
1
ℎ minus 120591119898(119905)
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904
lim120591119898(119905)rarr0
1
120591119898(119905)
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904 = 120578 (119905)
lim120591119898(119905)rarrℎ
1
ℎ minus 120591119898(119905)
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904 = 120578 (119905 minus ℎ)
(34)
From the Newton-Leibniz formula the following equa-tion is true for any matrices119872
119894119898 119873
119894119898 and 119881
119894119898with appro-
priate dimensions
(2120591119898(119905) 119880
⊺
1119872
119894119898+ 2 (ℎ minus 120591
119898(119905)) 119880
⊺
2119873
119894119898+ 2 120578
⊺
(119905) 119881119894119898)
times [minus 120578 (119905) + 119860119894120578 (119905) + 119861
120591119894120578 (119905 minus 120591
119898(119905)) + 119863
119894120596 (119905)] = 0
(35)
pound1198814(120578
119905 119894 119898)
le int
0
minusℎ
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904)
times ( sum
119899isinM119898 = 119899
120587119898119899119884119894119899
+ sum
119895isinN119894 = 119895
120582119894119895119884119895119898minus 119884)
times 120578 (119904) 119889119904 119889120579
+ 119890120575119905
120578⊺
(119905) 119884119894119898120578 (119905) int
0
minusℎ
119890minus120575120592
119889120592
minus 119890120575119905
int
119905
119905minusℎ
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
+ 119890120575119905
120578⊺
(119905) 119884 120578 (119905) int
0
minusℎ
int
0
120592
119890minus120575120592
119889120579 119889120592
(36)
Journal of Applied Mathematics 7
From Lemma 5 it yields that
minus int
119905
119905minusℎ
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
= minusint
119905
119905minus120591119898(119905)
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
minus int
119905minus120591119898(119905)
119905minusℎ
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
le minus
ℎ
120591119898(119905)
[int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904]
⊺
119884119894119898
ℎ
[int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904]
minus
ℎ
ℎ minus 120591119898(119905)
[int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904]
⊺
119884119894119898
ℎ
[int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904]
le minus
[
[
[
[
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904
]
]
]
]
⊺
[
[
[
119884119894119898
ℎ
119878119894119898
lowast
119884119894119898
ℎ
]
]
]
[
[
[
[
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904
]
]
]
]
(37)
From (25)ndash(37) we can eventually obtain
pound119881 (120578119905 119903
119905 119904
119905) minus 120575120596
⊺
(119905)119867120596 (119905) le 119890120575119905
120585⊺
(119905) Ξ119894119898120585 (119905) (38)
where
120585⊺
(119905) = [120578⊺
(119905) 120578⊺
(119905 minus 120591119898(119905)) 120578
⊺
(119905 minus ℎ) 120578⊺
(119905) 119880⊺
1 119880
⊺
2 120596
⊺
(119905)]
Ξ119894119898=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860⊺
119894119881
⊺
119894119898120591119898(119905) 119860
⊺
119894119872
⊺
119894119898(ℎ minus 120591
119898(119905)) 119860
⊺
119894119873
⊺
119894119898119875119894119898119863
119894
lowast Ξ22119894119898
119878119894119898minus
119884119894119898
ℎ
119860⊺
120591119894119881
⊺
119894119898120591119898(119905) 119860
⊺
120591119894119872
⊺
119894119898(ℎ minus 120591
119898(119905)) 119860
⊺
120591119894119873
⊺
1198941198980
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minus120591119898(119905)119872
⊺
119894119898minus (ℎ minus 120591
119898(119905))119873
⊺
119894119898119881119894119898119863
119894
lowast lowast lowast lowast minus120591119898(119905) 119883
1198941198980 120591
119898(119905)119872
119894119898119863
119894
lowast lowast lowast lowast lowast minus (ℎ minus 120591119898(119905))119883
119894119898(ℎ minus 120591
119898(119905))119873
119894119898119863
119894
lowast lowast lowast lowast lowast lowast minus120575119867
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
(39)
The LMIs (20) and (21) lead to 120591119898(119905) rarr ℎ and 120591
119898(119905) rarr
0 respectively It is easy to see that Ξ1119894119898
results from Ξ119894119898|
120591119898(119905) = ℎ and Ξ
119894119898|120591119898(119905) = 0 Thus we obtain
E pound119881 (120578119905 119903
119905 119904
119905) le 120578E [119881 (120578
119905 119903
119905 119904
119905)] + 120575120596
⊺
(119905)119867120596 (119905)
(40)
Multiplying the aforementioned inequality by 119890minus120578119905 we canget
E pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] le 119890
minus120578119905
120575120596⊺
(119905)119867120596 (119905) (41)
By integrating the aforementioned inequality between 0and 119905 it follows that
119890minus120578119905
E [119881 (120578119905 119903
119905 119904
119905)] minus E [119881 (120578
0 119903
0 119904
0)]
le 120575int
119905
0
119890minus120578119904
120596⊺
(119904)119867120596 (119904) 119889119904
(42)
Denote that 119894119898
= 119877minus(12)
119875119894119898119877minus(12) 119876
119904119894=
119877minus(12)
119876119904119894119877minus(12)
(119904 = 1 2) 119876 = 119877minus(12)
119876119877minus(12)
119883119894119898
= 119877minus(12)
119883119894119898119877minus(12) 119883 = 119877
minus(12)
119883119877minus(12)
119894119898
= 119877minus(12)
119884119894119898119877minus(12) and = 119877
minus(12)
119884119877minus(12) it yields
that
E [119881 (1205780 119903
0 119904
0)]
le max119894isinN119898isinM
120582max (119894119898) 120578⊺
(0) 119877120578 (0)
+ (max119894isinN
120582max (1198761119894) +max
119894isinN120582max (1198762119894
)) 119890120575ℎ
times int
0
minusℎ
119890120575119904
120578⊺
(119904) 119877120578 (119904) 119889119904
+ 119890120575ℎ
120582max (119876)int0
minusℎ
int
0
120579
119890120575119904
120578⊺
(119904) 119877120578 (119904) 119889119904
+ 119890120575ℎ max
119894isinN119898isinM120582max (119883119894119898
)
times int
0
minusℎ
int
0
120579
119890minus120575120579
120578⊺
(119904) 119877120578 (119904) 119889119904 119889120579
+ 119890120575ℎ
120582max (119883)
times int
0
minusℎ
int
0
120579
int
0
120592
119890minus120575120579
120578⊺
(119904) 119877120578 (119904) 119889119904 119889120579 119889120592
+ 119890120575ℎ max
119894isinN119898isinM120582max (119894119898)
8 Journal of Applied Mathematics
times int
0
minusℎ
int
0
120579
119890minus120575120579
120578⊺
(119904) 119877 120578 (119904) 119889119904 119889120579
+ 119890120575ℎ
120582max ()
times int
0
minusℎ
int
0
120579
int
0
120592
119890minus120575120579
120578⊺
(119904) 119877 120578 (119904) 119889119904 119889120579 119889120592
le max119894isinN119898isinM
120582max (119894119898)
+ ℎ119890120575ℎ
(max119894isinN
120582max (1198761119894+
119894isinNmax120582max (1198762119894
)))
+ ℎ2
119890120575ℎ
( max119894isinN119898isinM
120582max (119883119894119898) + 120582max (119876)
+ max119894isinN119898isinM
120582max (119894119898))
+
1
2
ℎ3
119890120575ℎ
(120582max (119883) + 120582max ())
times supminusℎle119904le0
120578⊺
(119904) 119877120578 (119904) 120578⊺
(119904) 119877 120578 (119904)
= 1198881Λ
(43)
For given 120578 gt 0 and 0 le 119905 le 119879 we have
E [119881 (120578119905 119903
119905 119904
119905)] le E [119890
120578119905
119881 (1205780 119903
0 119904
0)]
+ 119890120578119905
120575int
119905
0
119890minus120578119904
120596⊺
(119904)119867120596 (119904) 119889119904
le 119890120578119879
1198881Λ + 119889120575119890
120578119879
120582max (119867)int119879
0
119890minus120578119904
119889119904
le 119890120578119879
1198881Λ + 119889120575120582
10
1
120578
(1 minus 119890minus120578119879
)
(44)
On the other hand it follows from (25) that
E [119881 (120578119905 119903
119905 119904
119905)] ge E [120578
⊺
(119905) 119890120582119905
119875119894119898120578 (119905)]
ge max119894isinN
120582min (119875119894119898)E [120578⊺
(119905) 119877120578 (119905)]
= 1205821E [120578
⊺
(119905) 119877120578 (119905)]
(45)
It can be derived from (43)ndash(45) that
E [120578⊺
(119905) 119877120578 (119905)] le
119890120578119879
1205821
1198881Λ + 119889120575120582
10
1
120578
(1 minus 119890minus120578119879
) (46)
From (22) and (46) we have
E [120578⊺
(119905) 119877120578 (119905)] lt 1198882 (47)
Then the system is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879)
Remark 8 In this paper 120591119898(119905) and 120591
119898(119905) may have different
upper bounds in various delay intervals satisfying (7) respec-tively However in previous work such as [20 21] 120591
119898(119905) and
120591119898(119905) are enlarged to 120591
119898(119905) le ℎ = maxℎ
119894 119898 isin M and
120591119898(119905) le 120583 = max120583
119898 119898 isin M respectively which may lead
to conservativeness inevitably However the previouse casecan be taken fully into account by employing the Lyapunov-Krasovskii functional (25)
Remark 9 When dealing with termminusint
119905
119905minusℎ
120578⊺
(119904)119884119894119898120578(119904)119889119904 the
convex combination is not employed Lemma 5 is used in thispaper and then the free-weighting matrices-dependent nulladd items are necessary to be introduced in our proof whichlead to the decrease of the number of LMIs and LMIs scalardecision variables
Remark 10 The feature of this paper is the way to dealwith the integral term Many researchers have enlarged thederivative of the Lyapunov functional in order to deal with theintegral term in mathematical operations In this paper wepropose a novel delay-dependent sufficient criterion whichensures that the Markovian jump system with different modesystems is finite-time stable
Remark 11 It should be pointed out that the novelty of theLyapunov functional (25) lies in distinct Lyapunov matrices(119875
119894119898 119883
119894119898 119884
119894119898) which is chosen for different system modes
119894 (119894 = 1 2 119873) and119898 (119898 = 1 2 119872)
Theorem 12 System (9) is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879) if there exist matrices 119875
119894gt 0 119876
119897119894gt 0 (119897 = 1 2)
119876 gt 0 119883119894119898gt 0 119883 gt 0 119884
119894119898gt 0 119884 gt 0 119878
119894119898 scalars 119888
1lt 119888
2
119879 gt 0 120582119904gt 0 (119904 = 1 2 9) 120578 gt 0 120574 gt 0 and Λ gt 0 such
that for all 119894 119895 isinN and119898 119899 isinM and (16)ndash(19) the followinginequalities hold
Σ1119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119872⊺
119894119898119875119894119898119863119894
119862
⊺
119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119872⊺
1198941198980 119864
⊺
119894
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119872⊺
119894119898119881119894119898119863119894
0
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119872119894119898119863119894
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(48)
Σ2119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119873⊺
119894119898119875119894119898119863119894
119862
⊺
119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119873⊺
1198941198980 119864
⊺
119894
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119873⊺
119894119898119881119894119898119863119894
0
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119873119894119898119863119894
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(49)
1198881Λ + 119889120575120574
21
120578
(1 minus 119890minus120578119879
) lt 1205821119890minus120578119879
1198882 (50)
Journal of Applied Mathematics 9
Proof We now consider the 119867infin
performance of system (9)Select the same Lyapunov-Krasovskii functional as Theo-rem 7 it yields that
pound119881 (120578119905 119903
119905 119904
119905) + 119890
⊺
(119905) 119890 (119905) minus 1205742
120596⊺
(119905) 120596 (119905) le 120585⊺
(119905) Σ119897119894119898120585 (119905)
(119897 = 1 2)
(51)
It follows from (49)-(50) that
E pound119881 (120578119905 119903
119905 119904
119905)
le E [120578119881 (120578119905 119903
119905 119904
119905)] + 120574
2
120596⊺
(119905) 120596 (119905) minus E [119890⊺
(119905) 119890 (119905)]
(52)
Multiplying the aforementioned inequality by 119890minus120578119905 onehas
E pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] le 119890
minus120578119905
[1205742
120596⊺
(119905) 120596 (119905) minus 119890⊺
(119905) 119890 (119905)]
(53)
In zero initial condition and E[119881(120578119905 119903
119905 119904
119905)] gt 0 by inte-
grating the aforementioned inequality between 0 and 119879 wecan get
int
119879
0
119890minus120578119905
[1205742
120596⊺
(119905) 120596 (119905) minus 119890⊺
(119905) 119890 (119905)] 119889119905
le Eint119879
0
pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] 119889119905 le 119881 (120578
0 119903
0 119904
0) = 0
(54)
Using Dynkins formula it results that
E [int119879
0
119890minus120578120592
119890⊺
(120592) 119890 (120592) 119889120592] le 1205742
E [int119879
0
119890minus120578120592
120596⊺
(120592) 120596 (120592) 119889120592]
(55)
Then it yields that
E [int119879
0
119890⊺
(120592) 119890 (120592) 119889120592] le 1205742
119890120578119879
E [int119879
0
120596⊺
(120592) 120596 (120592) 119889120592] (56)
Thus it is concluded by Definition 3 that system (9) is fi-nite-time bounded with an 119867
infinperformance 120574 This com-
pletes the proof
Remark 13 From the proof process of Theorems 7 and 12it is easy to see that neither bounding technique for cross-terms nor model transformation is involved In other wordsthe obtained result is expected to be less conservative
Remark 14 The Lyapunov asymptotic stability and finite-time stability of a class of system are independent conceptsA Lyapunov asymptotic stability system may not be finite-time stability Moreover finite-time stability system may alsonot be Lyapunov asymptotic stabilityThere exist some resultson Lyapunov stability while finite-time stability also needsour full investigation which was neglected by most previouswork
4 Finite-Time 119867infin
Filtering
Theorem 15 System (9) is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879) if there exist matrices 119875
119894119898gt 0119872
119894119898 119873
119894119898 119881
119894119898
119876119897119894gt 0 (119897 = 1 2) 119876 gt 0 119883
119894119898gt 0 119883 gt 0 119884
119894119898gt 0 119884 gt 0
119860119891119894119898
119861119891119894119898
119862119891119894119898
119872119860119894119898
119872119861119894119898
119873119860119894119898
119873119861119894119898
119881119860119894119898
119881119861119894119898
119878119894119898
scalars 1198881lt 119888
2 119879 gt 0 120590
119904gt 0 (119904 = 1 2 9) 120575 gt 0 120578 gt 0
and Λ gt 0 such that for all 119894 119895 isin N and 119898 119899 isin M thefollowing inequalities hold
119875119894119898= [
1198751119894119898
1198752119894119898
1198752119894119898
1198752119894119898
] gt 0 119872119894119898= [
1198721119894119898
1198722119894119898
1198722119894119898
1198722119894119898
] (57)
119873119894119898= [
1198731119894119898
1198732119894119898
1198732119894119898
1198732119894119898
] 119881119894119898= [
1198811119894119898
1198812119894119898
1198812119894119898
1198812119894119898
] (58)
Γ1119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Γ11119894119898
Γ12119894119898
minus119878119894119898
Γ14119894119898
Γ15119894119898
Γ16119894119898
Γ17119894119898
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
Γ24119894119898
Γ25119894119898
0 [119862⊺
119911120591119894
0]
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119872⊺
119894119898Γ46119894119898
0
lowast lowast lowast lowast minusℎ119883119894119898
Γ56119894119898
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(59)
Γ2119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Γ11119894119898
Γ12119894119898
minus119878119894119898
Γ14119894119898
Γ1015840
15119894119898Γ16119894119898
Γ17119894119898
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
Γ24119894119898
Γ1015840
251198941198980 [
119862⊺
119911120591119894
0]
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119873⊺
119894119898Γ1015840
461198941198980
lowast lowast lowast lowast minusℎ119883119894119898
Γ1015840
561198941198980
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(60)where
Γ11119894119898
= sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898+ 119890
120575ℎ
1198761119894+ 119890
120575ℎ
1198762119894+ ℎ119876 +
119890120575ℎ
minus 1
120575
119883119894119898+
119890120575ℎ
minus 120575ℎ119890120575ℎ
minus 1
1205752
119883
+
119884119894119898
ℎ
+ [
1205751198751119894119898
+ 1198751119894119898119860
119894+ 119860
⊺
1198941198751119894119898
+ 119861119891119894119898119862119910119894+ 119862
⊺
119910119894119861
⊺
119891119894119898120575119875
2119894119898+ 119860
119891119894119898+ 119860
⊺
119891119894119898
1205751198752119894119898
+ 1198752119894119898119860
119894+ 119860
⊺
1198941198752119894119898
+ 119861119891119894119898119862119910119894+ 119862
⊺
119910119894119861
⊺
119891119894119898120575119875
2119894119898+ 119860
119891119894119898+ 119860
⊺
119891119894119898
]
(61)
10 Journal of Applied Mathematics
Γ12119894119898
= minus
119884119894119898
ℎ
+ 119878119894119898+ [
1198751119894119898119860
120591119894+ 119861
119891119894119898119862119910120591119894
1198752119894119898119860
120591119894+ 119861
119891119894119898119862119910120591119894
]
Γ14119894119898
= [
119860⊺
119894119881
⊺
1119894119898+ 119862
⊺
119910119894119881
⊺
119861119894119898119881
⊺
119860119894119898
119860⊺
119894119881
⊺
2119894119898+ 119862
⊺
119910119894119881
⊺
119861119894119898119881
⊺
119860119894119898
]
Γ15119894119898
= [
ℎ119860⊺
119894119872
⊺
1119894119898+ ℎ119862
⊺
119910119894119872
⊺
119861119894119898ℎ119872
⊺
119860119894119898
ℎ119860⊺
119894119872
⊺
2119894119898+ ℎ119862
⊺
119910119894119872
⊺
119861119894119898ℎ119872
⊺
119860119894119898
]
Γ1015840
15119894119898= [
ℎ119860⊺
119894119873
⊺
1119894119898+ ℎ119862
⊺
119910119894119873
⊺
119861119894119898ℎ119873
⊺
119860119894119898
ℎ119860⊺
119894119873
⊺
2119894119898+ ℎ119862
⊺
119910119894119873
⊺
119861119894119898ℎ119873
⊺
119860119894119898
]
Γ16119894119898
= [
1198751119894119898119863
119894+ 119861
119891119894119898119863
119910119894
1198752119894119898119863
119894+ 119861
119891119894119898119863
119910119894
]
Γ17119894119898
= [
119862⊺
119911119894
minus119862119891119894119898
]
Γ24119894119898
= [
119860⊺
120591119894119881
⊺
1119894119898+ 119862
⊺
119910120591119894119881
⊺
119861119894119898
119860⊺
120591119894119881
⊺
2119894119898+ 119862
⊺
119910120591119894119881
⊺
119861119894119898
]
Γ25119894119898
= [
ℎ119860⊺
120591119894119872
⊺
1119894119898+ ℎ119862
⊺
119910120591119894119872
⊺
119861119894119898
ℎ119860⊺
120591119894119872
⊺
2119894119898+ ℎ119862
⊺
119910120591119894119872
⊺
119861119894119898
]
Γ1015840
25119894119898= [
ℎ119860⊺
120591119894119873
⊺
1119894119898+ ℎ119862
⊺
119910120591119894119873
⊺
119861119894119898
ℎ119860⊺
120591119894119873
⊺
2119894119898+ ℎ119862
⊺
119910120591119894119873
⊺
119861119894119898
]
Γ46119894119898
= [
1198811119894119898119863
119894+ 119881
119861119894119898119863
119910119894
1198812119894119898119863
119894+ 119881
119861119894119898119863
119910119894
]
Γ56119894119898
= [
ℎ1198721119894119898119863
119894+ ℎ119872
119861119894119898119863
119910119894
ℎ1198722119894119898119863
119894+ ℎ119872
119861119894119898119863
119910119894
]
Γ1015840
56119894119898= [
ℎ1198731119894119898119863
119894+ ℎ119873
119861119894119898119863
119910119894
ℎ1198732119894119898119863
119894+ ℎ119873
119861119894119898119863
119910119894
]
(62)
Then a desired filter can be chosen with parameters as
119860119891119894119898
= 119875minus1
2119894119898119860
119891119894119898 119861
119891119894119898= 119875
minus1
2119894119898119861119891119894119898
119862119891119894119898
= 119862119891119894119898
(63)
Proof We denote that
119875119894119898= [
1198751119894119898
1198752119894119898
1198752119894119898
1198752119894119898
] (64)
The term 119875119894119898119860
119894can be rewritten as
119875119894119898119860
119894= [
1198751119894119898119860
119894+ 119875
21198941198981198611198911198941198981198621199101198941198752119894119898119860
119891119894119898
1198752119894119898119860
119894+ 119875
21198941198981198611198911198941198981198621199101198941198752119894119898119860
119891119894119898
] (65)
Similarly we have
119875119894119898119861119894= [
1198751119894119898119860
120591119894+ 119875
2119894119898119861119891119894119898119862119910120591119894
1198752119894119898119860
120591119894+ 119875
2119894119898119861119891119894119898119862119910120591119894
]
119875119894119898119863
119894= [
1198751119894119898119863
119894+ 119875
2119894119898119861119891119894119898119863
119910119894
1198752119894119898119863
119894+ 119875
2119894119898119861119891119894119898119863
119910119894
]
(66)
Define 119860119891119894119898
= 1198752119894119898119860
119891119894119898 119861
119891119894119898= 119875
2119894119898119861119891119894119898
119862119891119894119898
= 119862119891119894119898
119872
119860119894119898= 119872
2119894119898119860
119891119894119898119872
119861119894119898= 119872
2119894119898119861119891119894119898
119873119860119894119898
= 1198732119894119898119860
119891119894119898
119873119861119894119898
= 1198732119894119898119861119891119894119898
119881119860119894119898
= 1198812119894119898119860
119891119894119898 and 119881
119861119894119898= 119881
2119894119898119861119891119894119898
Therefore if (59) and (60) hold system (9) is finite-timebounded with a prescribed 119867
infinperformance index 120574 The
proof is completed
Remark 16 In many actual applications the minimum valueof 1205742min is of interest In Theorem 12 with a fixed 120582 120574min canbe obtained through the following optimization procedure
min 1205742
st (48)ndash(50) (67)
In Theorem 15 as for finite-time stability and bounded-ness once the state bound 119888
2is not ascertained theminimum
value 1198882min is of interest With a fixed 120582 define 120582
1= 1 then
the following optimization problem can be formulated to getminimum value 119888
2min
min 1205891205742
+ (1 minus 120589) 1198882
st (59)-(60) and (50)
(68)
where 120589 is weighted factor and 120589 isin [0 1]
5 Illustrative Example
Example 17 Consider that the Markovian jump system andthe delay mode switching are governed by a Markov processwith the following transition rates
Λ = [
minus07 07
09 minus09] Π = [
minus1 1
12 minus12] (69)
as well as with the following parameters
1198601= [
minus35 086
minus064 minus325] 119860
2= [
minus25 034
14 minus002]
1198601205911= [
minus08 minus13
minus07 minus22] 119860
1205912= [
minus28 05
minus08 minus14]
1198631= 119863
2= [
01
02] 119862
1199101= [04 07]
1198621199102= [05 08] 119862
1199101205911= [01 02]
1198621199101205912
= [01 01]
1198631199101= 119863
1199102= 01 119862
1199111= [01 005]
1198621199112= [02 009] 119862
1199111205911= [03 06]
1198621199111205912
= [04 08] 1198631199111= 119863
1199112= 005
(70)
Journal of Applied Mathematics 11
Then we choose 119877 = 119868 119879 = 2 1198881= 1 and 119889 = 001
throughTheorem 15 it yields that themode-dependent filtersare as follows
11986011989111
= [
minus61254 15565
minus02347 minus03763]
11986011989112
= [
minus62682 14338
minus03552 minus05327]
11986011989121
= [
minus89214 24223
minus15476 minus06234]
11986011989122
= [
minus84638 28237
minus13545 minus04322]
11986111989111
= [
51465 minus28516
minus91216 95171]
11986111989112
= [
54203 minus23121
minus93156 96332]
11986111989121
= [
46193 minus11984
minus162397 163006]
11986111989122
= [
46512 minus10311
minus161132 162312]
11986211989111
= [00294 minus00397]
11986211989112
= [00271 minus003368]
11986211989121
= [01163 minus01432]
11986211989122
= [01342 minus03672]
(71)
This paper deals with the finite-time filter design problemfor a class of Markovian jump systems particularly two dif-ferentMarkov processes are considered formodeling the ran-domness of system matrix and the state delay Then throughthe numerical example we can see that results in this paperare feasible which further verified the correctness of our the-ory Therefore the paper shorten this gap
6 Conclusions
In this paper we have examined the problems of finite-time119867
infinfiltering for a class of Markovian jump systems with dif-
ferent system modes Based on a novel approach a sufficientcondition is derived such that the closed-loop Markovianjump system is finite-time bounded and satisfies a prescribedlevel of119867
infindisturbance attenuation in a finite time interval
Finally a numerical example is also given to illustrate theeffectiveness of the proposed design approach It should benoted that one of future research topics would be to investi-gate the problems of fault detection and fault tolerant controlfor time-varying Markovian jump systems with incompleteinformation over a finite-time horizon
Acknowledgments
The authors would like to thank the associate editor and theanonymous reviewers for their detailed comments and sug-gestions This work was supported by the Fund of SichuanProvincial Key Laboratory of Signal and Information Pro-cessing Xihua University (SZJJ2009-002 and SGXZD0101-10-1) and National Basic Research Program of China(2010CB732501)
References
[1] N N Krasovskiı and E A Lidskiı ldquoAnalytical design ofcontrollers in systems with random attributesmdashI Statementof the problem method of solvingrdquo Automation and RemoteControl vol 22 pp 1021ndash1025 1961
[2] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[3] L Wu X Su and P Shi ldquoSliding mode control with bounded1198712gain performance of Markovian jump singular time-delay
systemsrdquo Automatica vol 48 no 8 pp 1929ndash1933 2012[4] H Gao Z Fei J Lam and B Du ldquoFurther results on expo-
nential estimates of Markovian jump systems with mode-de-pendent time-varying delaysrdquo IEEE Transactions on AutomaticControl vol 56 no 1 pp 223ndash229 2011
[5] Z Wu H Su and J Chu ldquo119867infinfiltering for singular Markovian
jump systems with time delayrdquo International Journal of Robustand Nonlinear Control vol 20 no 8 pp 939ndash957 2010
[6] P Tino M Cernansky and L Benuskova ldquoMarkovian architec-tural bias of recurrent neural networksrdquo IEEE Transactions onNeural Networks vol 15 no 1 pp 6ndash15 2004
[7] H Dong Z Wang D W C Ho and H Gao ldquoRobust 119867infin
filtering for Markovian jump systems with randomly occurringnonlinearities and sensor saturation the finite-horizon caserdquoIEEE Transactions on Signal Processing vol 59 no 7 pp 3048ndash3057 2011
[8] Q Ma S Xu and Y Zou ldquoStability and synchronization forMarkovian jump neural networks with partly unknown tran-sition probabilitiesrdquo Neurocomputing vol 74 pp 3404ndash33412011
[9] H Zhao S Xu and Y Zou ldquoRobust119867infinfiltering for uncertain
Markovian jump systems with mode-dependent distributeddelaysrdquo International Journal of Adaptive Control and SignalProcessing vol 24 no 1 pp 83ndash94 2010
[10] Y Zhang S Xu and B Zhang ldquoDiscrete-time fuzzy Markovianjump systems with time-varying delaysrdquo IEEE Transactions onFuzzy Systems vol 17 no 2 pp 411ndash420 2009
[11] Z Wu H Su and J Chu ldquoState estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 no 10ndash12 pp 2247ndash2254 2010
[12] Q Zhu and J Cao ldquoRobust exponential stability of markovianjump impulsive stochastic Cohen-Grossberg neural networkswithmixed time delaysrdquo IEEE Transactions onNeural Networksvol 21 no 8 pp 1314ndash1325 2010
[13] H Huang G Feng and X Chen ldquoStability and stabilizationof Markovian jump systems with time delay via new Lyapunovfunctionalsrdquo IEEE Transactions on Circuits and Systems I vol59 no 10 pp 2413ndash2421 2012
12 Journal of Applied Mathematics
[14] Q Zhu and J Cao ldquoStability of Markovian jump neural net-works with impulse control and time varying delaysrdquoNonlinearAnalysis Real World Applications vol 13 no 5 pp 2259ndash22702012
[15] M Sun J Lam S Xu andY Zou ldquoRobust exponential stabiliza-tion for Markovian jump systems with mode-dependent inputdelayrdquo Automatica vol 43 no 10 pp 1799ndash1807 2007
[16] M Mohammadian A Abolmasoumi and H Momeni ldquo119867infin
mode-independent filter design for Markovian jump geneticregulatory networks with time-varying delaysrdquo Neurocomput-ing vol 87 pp 10ndash18 2012
[17] S Boyd L El Ghaoui E Feron and V Balakrishnan LinearMatrix Inequalities in SystemandControlTheory vol 15 of SIAMStudies in Applied Mathematics SIAM Philadelphia Pa USA1994
[18] S Mou H GaoW Qiang and K Chen ldquoNew delay-dependentexponential stability for neural networks with time delayrdquo IEEETransactions on Systems Man and Cybernetics Part B vol 38no 2 pp 571ndash576 2008
[19] O M Kwon J Park S Lee and E Cha ldquoAnalysis on delay-dependent stability for neural networks with time-varyingdelaysrdquo Neurocomputing vol 103 pp 114ndash120 2013
[20] L Hu H Gao andW X Zheng ldquoNovel stability of cellular neu-ral networkswith interval time-varying delayrdquoNeuralNetworksvol 21 no 10 pp 1458ndash1463 2008
[21] H Zhang Z Liu G B Huang and ZWang ldquoNovel weighting-delay-based stability criteria for recurrent neural networks withtime-varying delayrdquo IEEE Transactions on Neural Networks vol21 no 1 pp 91ndash106 2010
[22] Z Zuo C Yang and YWang ldquoA newmethod for stability anal-ysis of recurrent neural networks with interval time-varyingdelayrdquo IEEE Transactions on Neural Networks vol 21 no 2 pp339ndash344 2010
[23] P Park JWKo andC Jeong ldquoReciprocally convex approach tostability of systems with time-varying delaysrdquo Automatica vol47 no 1 pp 235ndash238 2011
[24] D Zhang L Yu and W Zhang ldquoDelay-dependent fault detec-tion for switched linear systems with time-varying delays-theaverage dwell time approachrdquo Signal Processing vol 91 pp 832ndash840 2011
[25] Z Fei H Gao and P Shi ldquoNew results on stabilization ofMark-ovian jump systems with time delayrdquoAutomatica vol 45 no 10pp 2300ndash2306 2009
[26] H R Karimi ldquoRobust delay-dependent 119867infin
control of uncer-tain time-delay systems with mixed neutral discrete anddistributed time-delays and Markovian switching parametersrdquoIEEE Transactions on Circuits and Systems I vol 58 no 8 pp1910ndash1923 2011
[27] M Luo G Liu and S Zhong ldquoRobust fault detection ofMarkovian jump systems with different systemmodesrdquoAppliedMathematical Modelling vol 37 no 7 pp 5001ndash5012 2013
[28] B Zhang andY Li ldquoExponential1198712minus119871
infinfiltering for distributed
delay systems with Markovian jumping parametersrdquo SignalProcessing vol 93 no 1 pp 206ndash216 2013
[29] A Abolmasoumi and H Momeni ldquoRobust observer-based119867infin
control of a Markovian jump system with different delay andsystem modesrdquo International Journal of Control Automationand Systems vol 9 no 4 pp 768ndash776 2011
[30] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[31] S Ma and E K Boukas ldquoRobust 119867infin
filtering for uncertaindiscrete Markov jump singular systems with mode-dependenttime delayrdquo IET Control Theory amp Applications vol 3 no 3 pp351ndash361 2009
[32] Y ChenW Bi andW Li ldquoStability analysis for neural networkswith time-varying delay a more general delay decompositionapproachrdquo Neurocomputing vol 73 no 4ndash6 pp 853ndash857 2010
[33] Z Shu J Lam and S Xu ldquoRobust stabilization of Markoviandelay systems with delay-dependent exponential estimatesrdquoAutomatica vol 42 no 11 pp 2001ndash2008 2006
[34] X Huang W Lin and B Yang ldquoGlobal finite-time stabilizationof a class of uncertain nonlinear systemsrdquo Automatica vol 41no 5 pp 881ndash888 2005
[35] C Qian and J Li ldquoGlobal finite-time stabilization by outputfeedback for planar systems without observable linearizationrdquoIEEE Transactions on Automatic Control vol 50 no 6 pp 885ndash890 2005
[36] F Amato R Ambrosino C Cosentino and G De TommasildquoInput-output finite time stabilization of linear systemsrdquo Auto-matica vol 46 no 9 pp 1558ndash1562 2010
[37] Y Hong ldquoFinite-time stabilization and stabilizability of a classof controllable systemsrdquo Systems amp Control Letters vol 46 no4 pp 231ndash236 2002
[38] F Amato and M Ariola ldquoFinite-time control of discrete-timelinear systemsrdquo IEEETransactions onAutomatic Control vol 50no 5 pp 724ndash729 2005
[39] Z Zuo Y Liu Y Wang and H Li ldquoFinite-time stochasticstability and stabilisation of linear discrete-time Markovianjump systems with partly unknown transition probabilitiesrdquoIET Control Theory amp Applications vol 6 no 10 pp 1522ndash15262012
[40] X Luan F Liu and P Shi ldquoFinite-time filtering for non-linearstochastic systems with partially known transition jump ratesrdquoIET Control Theory amp Applications vol 4 no 5 pp 735ndash7452010
[41] F Amato M Ariola and C Cosentino ldquoFinite-time control ofdiscrete-time linear systems analysis and design conditionsrdquoAutomatica vol 46 no 5 pp 919ndash924 2010
[42] X Lin H Du and S Li ldquoFinite-time boundedness and 1198712-
gain analysis for switched delay systems with norm-boundeddisturbancerdquo Applied Mathematics and Computation vol 217no 12 pp 5982ndash5993 2011
[43] J Lin S Fei and Z Gao ldquoStabilization of discrete-time switchedsingular time-delay systems under asynchronous switchingrdquoJournal of the Franklin Institute vol 349 no 5 pp 1808ndash18272012
[44] H Song L Yu D Zhang and W-A Zhang ldquoFinite-time 119867infin
control for a class of discrete-time switched time-delay systemswith quantized feedbackrdquoCommunications in Nonlinear Scienceand Numerical Simulation vol 17 no 12 pp 4802ndash4814 2012
[45] Z Zuo H Li Y Liu and Y Wang ldquoOn finite-time stochasticstability and stabilization ofMarkovian jump systems subject topartial information on transition probabilitiesrdquoCircuits Systemsand Signal Processing 2012
[46] H Liu Y Shen and X Zhao ldquoDelay-dependent observer-based119867
infinfinite-time control for switched systems with time-varying
delayrdquoNonlinear Analysis Hybrid Systems vol 6 no 3 pp 885ndash898 2012
[47] Y Zhang C Liu and X Mu ldquoRobust finite-time119867infincontrol of
singular stochastic systems via static output feedbackrdquo AppliedMathematics and Computation vol 218 no 9 pp 5629ndash56402012
Journal of Applied Mathematics 13
[48] J Cheng H Zhu S Zhong Y Zhang and Y Zeng ldquoFinite-timeStabilization of 119867
infinfiltering for switched stochastic systemsrdquo
Circuits Systems and Signal Processing 2012[49] J Cheng H Zhu S Zhong and Y Zhang ldquoFinite-time bound-
ness of 119867infin
filtering for switching discrete-time systemsrdquo In-ternational Journal of Control Automation and Systems vol 10pp 1129ndash1135 2012
[50] Z Xiang C Qiao and M S Mahmoud ldquoFinite-time analysisand119867
infincontrol for switched stochastic systemsrdquo Journal of the
Franklin Institute vol 349 no 3 pp 915ndash927 2012
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Journal of Applied Mathematics 3
matrices with appropriate dimensions functions of the ran-dom jumping process 119903
119905 and represent the nominal systems
for each 119903119905isin N For notation simplicity when the system
operates in the 119894-th mode (119903119905= 119894) 119860
119903119905
119860120591119903119905
119863119903119905
119862119910119903119905
119862119910120591119903119905
119863
119910119903119905
119862119911119903119905
119862119911120591119903119905
and119863119911119903119905
are denoted as119860119894119860
120591119894119863
119894119862
119910119894119862
119910120591119894
119863119910119894 119862
119911119894 119862
119911120591119894 and119863
119911119894 respectively
Here we are interested in designing a full-order filterdescribed by
119891(119905) = 119860
119891119903119905119904119905
119909119891(119905) + 119861
119891119903119905119904119905
119910 (119905)
119911119891(119905) = 119862
119891119903119905119904119905
119909119891(119905)
(8)
where 119909119891(119905) isin R119899 is the filter state 119911
119891(119896) isin R119902 and the
matrices 119860119891119903119905119904119905
119861119891119903119905119904119905
and 119862119891119903119905119904119905
are unknown filter param-eters to be designed
Augmenting the model of (1) to include the filter (8) weobtain the following filtering error system
120578 (119905) = 119860119903119905
120578 (119905) + 119861119903119905
120578 (119905 minus 120591119898(119905)) + 119863
119903119905
120596 (119905)
119890 (119905) = 119862119903119905
120578 (119905) + 119864119903119905
120578 (119905 minus 120591119898(119905))
(9)
where
120578 (119905) = [
119909 (119905)
119909119891(119905)] 119890 (119905) = 119911 (119905) minus 119911
119891(119905)
119860119903119905
= [
119860119903119905
0
119861119891119903119905119904119905
119862119910119903119905
119860119891119903119905119904119905
] 119861119903119905
= [
119860120591119903119905
119861119891119903119905119904119905
119862119910120591119903119905
]
119863119903119905
= [
119863119903119905
119861119891119903119905119904119905
119863119910119903119905
]
119862119903119905
= [119862119911119903119905
minus119862119891119903119905119904119905
] 119864119903119905
= [119862119911120591119903119905
0]
(10)
In order to more precisely describe the main objectivewe introduce the following definitions and Lemmas for theunderlying system
Definition 1 System (1) is said to be finite-time bounded withrespect to (119888
1 119888
2 119879 119877 119889) if condition (2) and the following
inequality hold
supminusℎle120592le0
E 120578⊺
(120592) 119877120578 (120592) 120578⊺
(120592) 119877 120578 (120592)
le 1198881997904rArr E 120578
⊺
(119905) 119877120578 (119905) lt 1198882
forall119905 isin [0 119879]
(11)
where 1198882gt 119888
1ge 0 and 119877 gt 0
Definition 2 Consider 119881(120578119905 119903
119905 119904
119905 119905 gt 0) as the stochastic
positive Lyapunov function its weak infinitesimal operatoris defined as
pound119881 (120578119905 119903
119905 119904
119905)
= limΔrarr0
1
Δ
[E 119881 (120578119905+Δ 119903
119905+Δ 119904
119905+Δ) | 120578
119905 119903
119905 119904
119905
minus119881 (120578119905 119903
119905 119904
119905)]
(12)
Definition 3 Given a constant 119879 gt 0 for all admissible120596(119905) subject to condition (2) under zero initial conditionsif the closed-loop Markovian jump system (1) is finite-timebounded and the control outputs satisfy condition (8) withattenuation 120574 gt 0
Eint119879
0
119890⊺
(119905) 119890 (119905) 119889119905 le 1205742
119890120578119879
Eint119879
0
120596⊺
(119905) 120596 (119905) 119889119905 (13)
Then the controller system (1) finite-time bounded withdisturbance attenuation 120574
Remark 4 It should be pointed out that the assumption ofzero initial condition in system (1) is only for the purposeof technical simplification in the derivation and it does notcause loss of generality In fact if this assumption is lost thesame control result can be obtained along the same lineexcept for adding extra manipulations in the derivation andextra terms in the control presentation However in real-world applications the initial condition of the underlyingsystem is generally not zero
Lemma 5 (see [32]) Let 119891119894 R119898
rarr R (119894 = 1 2 119873)
have positive values in an open subset D of R119898 Then the re-ciprocally convex combination of 119891
119894overD satisfies
min120573119894|120573119894gt0sum119894120573119894=1
sum
119894
1
120573119894
119891119894(119905) = sum
119894
119891119894(119905) +max
119892119894119895(119905)
sum
119894 = 119895
119892119894119895(119905)
119904119906119887119895119890119888119905 119905119900 119892119894119895R
119898
997888rarrR 119892119895119894(119905) = 119892
119894119895(119905)
[
119891119894(119905) 119892
119894119895(119905)
119892119894119895(119905) 119891
119895(119905)] ge 0
(14)
Lemma 6 (Schur Complement [17]) Given constant matrices119883119884119885 where119883 = 119883
⊺ and 0 lt 119884 = 119884⊺ then119883+119885⊺
119884minus1
119885 lt 0
if and only if
[
119883 119885⊺
lowast minus119884] lt 0 119900119903 [
minus119884 119885
lowast 119883] lt 0 (15)
3 Finite-Time 119867infin
Performance Analysis
Theorem 7 System (9) is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879) if there exist matrices 119875
119894119898gt 0 119876
119897119894gt 0 (119897 = 1
2) 119876 gt 0 119883119894119898gt 0 119883 gt 0 119884
119894119898gt 0 119884 gt 0119867 gt 0 119878
119894119898 scalars
1198881lt 119888
2 119879 gt 0120582
119904gt 0 (119904 = 1 2 10) 120578 gt 0 and Λ gt 0 such
that for all 119894 119895 isin N and 119898 119899 isin M the following inequalitieshold
sum
119895isinN119895 = 119894
120582119894119895(119876
1119895+ 119876
2119895) + sum
119899isinM119898 = 119899
120587119898119899119876
2119894minus 119876 lt 0 (16)
sum
119899isinM119898 = 119899
120587119898119899119883
119894119899+ sum
119895isinN119894 = 119895
120582119894119895119883
119895119898minus 119883 lt 0 (17)
4 Journal of Applied Mathematics
sum
119899isinM119898 = 119899
120587119898119899119884119894119899+ sum
119895isinN119894 = 119895
120582119894119895119884119895119898minus 119884 lt 0 (18)
[
[
[
119884119894119898
ℎ
119878119894119898
lowast
119884119894119898
ℎ
]
]
]
gt 0 (19)
Ξ1119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119872⊺
119894119898119875119894119898119863119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119872⊺
1198941198980
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119872⊺
119894119898119881119894119898119863119894
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119872119894119898119863119894
lowast lowast lowast lowast lowast minus120575119867
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(20)
Ξ2119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119872⊺
119894119898119875119894119898119863119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119873⊺
1198941198980
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119873⊺
119894119898119881119894119898119863119894
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119873119894119898119863119894
lowast lowast lowast lowast lowast minus120575119867
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(21)
1198881Λ + 119889120575120582
10
1
120578
(1 minus 119890minus120578119879
) lt 1205821119890minus120578119879
1198882 (22)
where
Ξ11119894119898
= sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898
+ 120575119875119894119898+ 119875
119894119898119860
119894+ 119860
⊺
119894119875119894119898+ 119890
120575ℎ
1198761119894
+ 119890120575ℎ
1198762119894+ ℎ119876
119890120575ℎ
minus 1
120575
119883119894119898
+
119890120575ℎ
minus 120575ℎ119890120575ℎ
minus 1
1205752
119883 +
119884119894119898
ℎ
Ξ12119894119898
= 119875119894119898119861119894minus
119884119894119898
ℎ
+ 119878119894119898
Ξ22119894119898
= 119903 (120583119898) 119876
2119894+
2119884119894119898
ℎ
minus 119878119894119898minus 119878
⊺
119894119898
Ξ44119894119898
= minus119881119894119898minus 119881
⊺
119894119898+
119890120575ℎ
minus 1
120575
119884119894119898+
119890120575ℎ
minus 120575ℎ119890120575ℎ
minus 1
1205752
119884
119903 (120591119898) =
minus (1 minus 120583119898) 119890
120575ℎ119894 if 120583
119898gt 1
minus (1 minus 120583119898) if 120583
119898le 1
Λ = 1205822+ ℎ119890
120575ℎ
(1205823+ 120582
4) + ℎ
2
119890120575ℎ
(1205825+ 120582
6+ 120582
8)
+
1
2
ℎ3
119890120575ℎ
(1205827+ 120582
9)
1205821= min
119894isinN119898isinM120582min (119894119898) 120582
2= max
119894isinN119898isinM120582max (119894119898)
1205823= max
119894isinN120582max (1198761119894
)
1205824= max
119894isinN120582max (1198762119894
) 1205825= 120582max (119876)
1205826= max
119894isinN119898isinM120582max (119883119894119898
) 1205827= 120582max (119883)
1205828= max
119894isinN119898isinM120582max (119894119898) 120582
9= 120582max ()
12058210= 120582max (119867)
119894119898= 119877
minus(12)
119875119894119898119877minus(12)
119876119897119894= 119877
minus(12)
119876119897119894119877minus(12)
(119897 = 1 2)
119876 = 119877minus(12)
119876119877minus(12)
119883119894119898= 119877
minus(12)
119883119894119898119877minus(12)
119883 = 119877minus(12)
119883119877minus(12)
119894119898= 119877
minus(12)
119884119894119898119877minus(12)
= 119877minus(12)
119884119877minus(12)
(23)
Proof First in order to cast our model into the framework oftheMarkov processes we define a new process (120578
119905 119903
119905 119904
119905) 119905 ge
0 by
120578119905(119904) = 120578 (119905 + 119904) 119904 isin [minusℎ 0] (24)
Now we consider the following Lyapunov-Krasovskiifunctional
119881 (120578119905 119903
119905 119904
119905) =
4
sum
119897=1
119881119897(120578
119905 119903
119905 119905) (25)
where
1198811(120578
119905 119903
119905 119904
119905) = 120578(119905)
⊺
119890120575119905
119875119903119905119904119905
120578 (119905)
1198812(120578
119905 119903
119905 119904
119905) = int
119905
119905minusℎ
119890120575(119904+ℎ)
120578⊺
(119904) 1198761119903119905
120578 (119904) 119889119904
+ int
119905
119905minus120591119904119905(119905)
119890120575(119904+ℎ)
120578⊺
(119904) 1198762119903119905
120578 (119904) 119889119904
+ int
0
minusℎ
int
119905
119905+120579
119890120575(119904+ℎ)
120578⊺
(119904) 119876120578 (119904) 119889119904 119889120579
1198813(120578
119905 119903
119905 119904
119905) = int
0
minusℎ
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904)119883119903119905119904119905
120578 (119904) 119889119904 119889120579
+ int
0
minusℎ
int
0
120579
int
119905
119905+120592
119890120575(119904minus120579)
120578⊺
(119904)119883120578 (119904) 119889119904 119889120592 119889120579
1198814(120578
119905 119903
119905 119904
119905) = int
0
minusℎ
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904) 119884119903119905119904119905
120578 (119904) 119889119904 119889120579
+ int
0
minusℎ
int
0
120579
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904) 119884 120578 (119904) 119889119904 119889120592 119889120579
(26)
Journal of Applied Mathematics 5
Then for each 119903119905= 119894 119904
119905= 119898 we have
pound1198811(120578
119905 119894 119898)
= limΔrarr0
+
1
Δ
E 120578⊺
(119905 + Δ) 119890120575(119905+Δ)
119875119903119905+Δ
119904119905+Δ
120578 (119905 + Δ)
minus120578⊺
(119905) 119890120575119905
119875119894119898120578 (119905)
= limΔrarr0
+
1
Δ
E
120578⊺
(119905 + Δ)[
[
(1 + 120582119894119894Δ + 119900 (Δ))
times (1 + 120587119898119898Δ + 119900 (Δ)) 119890
120575(119905+Δ)
119875119894119898
+ (1 + 120582119894119894Δ + 119900 (Δ))
times ( sum
119899isinM
(120587119898119899Δ + 119900 (Δ))) 119890
120575(119905+Δ)
119875119894119899
+ (1 + 120587119898119898Δ + 119900 (Δ))
times (sum
119895isinN
(120582119894119895Δ + 119900 (Δ))) 119890
120575(119905+Δ)
119875119895119898
+ ( sum
119899isinM
(120587119898119899Δ + 119900 (Δ)))
times(sum
119895isinN
(120582119894119895Δ + 119900 (Δ))) 119890
120575(119905+Δ)
119875119895119899
]
]
times 120578 (119905 + Δ)
minus 120578⊺
(119905) 119890120575119905
119875119894119898120578 (119905)
= 120575119890120575119905
120578⊺
(119905) 119875119894119898120578 (119905) + 2119890
120575119905
120578⊺
(119905) 119875119894119898120578 (119905)
+ 119890120575119905
120578⊺
(119905)( sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898)120578 (119905)
= 119890120575119905
120578⊺
(119905)( sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898+ 120575119875
119894119898
+ 119875119894119898119860
119894+ 119860
⊺
119894119875119894119898)120578 (119905)
+ 2119890120575119905
120578⊺
(119905) 119875119894119898119861119894120578 (119905 minus 120591
119898(119905)) + 2119890
120575119905
120578⊺
(119905) 119875119894119898119863
119894120596 (119905)
(27)pound119881
2(120578
119905 119894 119898)
= limΔrarr0
+
1
Δ
Eint119905+Δ
119905+Δminusℎ
120578⊺
(119904) 119890120575(119904+ℎ)
1198761119903119905+Δ
120578 (119904) 119889119904
minusint
119905
119905minusℎ
120578⊺
(119904) 119890120575(119904+ℎ)
1198761119894120578 (119904) 119889119904
+ limΔrarr0
+
1
Δ
Eint119905+Δ
119905+Δminus120591119904119905+Δ
(119905)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119903119905+Δ
120578 (119904) 119889119904
minusint
119905
119905minus120591119898(119905)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119894120578 (119904) 119889119904
+ limΔrarr0
+
1
Δ
Eint0
minusℎ
int
119905+Δ
119905+Δ+120579
119890120575(119904+ℎ)
120578⊺
(119904) 119876120578 (119904) 119889119904 119889120579
minusint
0
minusℎ
int
119905
119905+120579
119890120575(119904+ℎ)
120578⊺
(119904) 119876120578 (119904) 119889119904 119889120579
= limΔrarr0
+
1
Δ
Eint119905+Δ
119905+Δminusℎ
120578⊺
(119904) 119890120575(119904+ℎ)
1198761119894120578 (119904) 119889119904
+ int
119905+Δ
119905+Δminusℎ
120578⊺
(119904)
times sum
119895isinN
(120582119894119895Δ + 119900 (Δ))
times 119890120575(119904+ℎ)
1198761119894120578 (119904) 119889119904
minusint
119905
119905minusℎ
120578⊺
(119904) 1198761119894120578 (119904) 119889119904
+ limΔrarr0
+
1
Δ
E[int119905+Δ
119905+Δminus120591119895(119905+Δ)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119894120578 (119904) 119889119904
+ sum
119899isinM
(120587119898119899Δ + 119900 (Δ))
times int
119905
119905+Δminus120591119899(119905+Δ)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119894120578 (119904) 119889119904
+ int
119905+Δ
minus120591119895(119905 + Δ)
119905+Δ
120578⊺
(119904)
times sum
119895isinN
(120582119894119895Δ + 119900 (Δ))
times 119890120575(119904+ℎ)
1198762119895120578 (119904) 119889119904
minusint
119905
119905minus120591119898(119905)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119894120578 (119904) 119889119904]
+ 119890120575119905
120578⊺
(119905) ℎ119876120578 (119905) minus int
119905minusℎ
119890120575(119904+ℎ)
120578⊺
(119904) 119876120578 (119904) 119889119904
le 119890120575119905
120578⊺
(119905) (119890120575ℎ
1198761119894+ 119890
120575ℎ
1198762119894+ ℎ119876) 120578 (119905)
minus 119890120575119905
120578⊺
(119905 minus ℎ)1198761119894120578 (119905 minus ℎ)
minus (1 minus 120591119898(119905)) 119890
120575119905
120578⊺
(119905 minus 120591119898(119905)) 119876
2119894120578 (119905 minus 120591
119898(119905))
+ int
119905
119905minusℎ
119890120590(119904+ℎ)
120578⊺
(119904)( sum
119895isinN119895 = 119894
120582119894119895(119876
1119895+ 119876
2119895)
6 Journal of Applied Mathematics
+ sum
119899isinM119898 = 119899
120587119898119899119876
2119894minus 119876)
times 120578 (119904) 119889119904
(28)
Since
0 le 120591119898(119905) le ℎ
119898 (29)
we define
119903 (120591119898) =
minus (1 minus 120583119898) 119890
120575ℎ119898 if 120583
119898gt 1
minus (1 minus 120583119898) if 120583
119898le 1
(30)
Then
pound1198812(120578
119905 119894 119898) le 119890
120575119905
120578⊺
(119905) (119890120575ℎ
1198761119894+ 119890
120575ℎ
1198762119894+ ℎ119876) 120578 (119905)
minus 119890120575119905
120578⊺
(119905 minus ℎ)1198761119894120578 (119905 minus ℎ)
+ 119890120575119905
120578⊺
(119905 minus 120591119898(119905)) 119903 (120591
119898) 119876
2119894120578 (119905 minus 120591
119898(119905))
+ int
119905
119905minusℎ
119890120590(119904+ℎ)
120578⊺
(119904)
times ( sum
119895isinN119895 = 119894
120582119894119895(119876
1119895+ 119876
2119895)
+ sum
119899isinM119898 = 119899
120587119898119899119876
2119894minus 119876)
times 120578 (119904) 119889119904
(31)
Similar to the previous process we can obtain
pound1198813(120578
119905 119894 119898) le int
0
minusℎ
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904)
times ( sum
119899isinM119898 = 119899
120587119898119899119883
119894119899
+ sum
119895isinN119894 = 119895
120582119894119895119883
119895119898minus 119883)
times 120578 (119904) 119889119904 119889120579
+ 119890120575119905
120578⊺
(119905) 119883119894119898120578 (119905) int
0
minusℎ
119890minus120575120592
119889120592
minus 119890120575119905
int
119905
119905minusℎ
120578⊺
(119904)119883119894119898120578 (119904) 119889119904
+ 119890120575119905
120578⊺
(119905) 119883120578 (119905) int
0
minusℎ
int
0
120592
119890minus120575120592
119889120579119889120592
(32)
By using Lemma 5 it yields that
minusint
119905
119905minusℎ
120578⊺
(119904)119883119894119898120578 (119904) 119889119904 = minusint
119905
119905minus120591119898(119905)
120578⊺
(119904)119883119894119898120578 (119904) 119889119904
minus int
119905minus120591119898(119905)
119905minusℎ
120578⊺
(119904)119883119894119898120578 (119904) 119889119904
le minus120591119898(119905) 119880
⊺
1119883
1198941198981198801
minus (ℎ minus 120591119898(119905)) 119880
⊺
2119883
1198941198981198802
(33)
where
1198801=
1
120591119898(119905)
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904
1198802=
1
ℎ minus 120591119898(119905)
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904
lim120591119898(119905)rarr0
1
120591119898(119905)
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904 = 120578 (119905)
lim120591119898(119905)rarrℎ
1
ℎ minus 120591119898(119905)
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904 = 120578 (119905 minus ℎ)
(34)
From the Newton-Leibniz formula the following equa-tion is true for any matrices119872
119894119898 119873
119894119898 and 119881
119894119898with appro-
priate dimensions
(2120591119898(119905) 119880
⊺
1119872
119894119898+ 2 (ℎ minus 120591
119898(119905)) 119880
⊺
2119873
119894119898+ 2 120578
⊺
(119905) 119881119894119898)
times [minus 120578 (119905) + 119860119894120578 (119905) + 119861
120591119894120578 (119905 minus 120591
119898(119905)) + 119863
119894120596 (119905)] = 0
(35)
pound1198814(120578
119905 119894 119898)
le int
0
minusℎ
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904)
times ( sum
119899isinM119898 = 119899
120587119898119899119884119894119899
+ sum
119895isinN119894 = 119895
120582119894119895119884119895119898minus 119884)
times 120578 (119904) 119889119904 119889120579
+ 119890120575119905
120578⊺
(119905) 119884119894119898120578 (119905) int
0
minusℎ
119890minus120575120592
119889120592
minus 119890120575119905
int
119905
119905minusℎ
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
+ 119890120575119905
120578⊺
(119905) 119884 120578 (119905) int
0
minusℎ
int
0
120592
119890minus120575120592
119889120579 119889120592
(36)
Journal of Applied Mathematics 7
From Lemma 5 it yields that
minus int
119905
119905minusℎ
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
= minusint
119905
119905minus120591119898(119905)
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
minus int
119905minus120591119898(119905)
119905minusℎ
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
le minus
ℎ
120591119898(119905)
[int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904]
⊺
119884119894119898
ℎ
[int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904]
minus
ℎ
ℎ minus 120591119898(119905)
[int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904]
⊺
119884119894119898
ℎ
[int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904]
le minus
[
[
[
[
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904
]
]
]
]
⊺
[
[
[
119884119894119898
ℎ
119878119894119898
lowast
119884119894119898
ℎ
]
]
]
[
[
[
[
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904
]
]
]
]
(37)
From (25)ndash(37) we can eventually obtain
pound119881 (120578119905 119903
119905 119904
119905) minus 120575120596
⊺
(119905)119867120596 (119905) le 119890120575119905
120585⊺
(119905) Ξ119894119898120585 (119905) (38)
where
120585⊺
(119905) = [120578⊺
(119905) 120578⊺
(119905 minus 120591119898(119905)) 120578
⊺
(119905 minus ℎ) 120578⊺
(119905) 119880⊺
1 119880
⊺
2 120596
⊺
(119905)]
Ξ119894119898=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860⊺
119894119881
⊺
119894119898120591119898(119905) 119860
⊺
119894119872
⊺
119894119898(ℎ minus 120591
119898(119905)) 119860
⊺
119894119873
⊺
119894119898119875119894119898119863
119894
lowast Ξ22119894119898
119878119894119898minus
119884119894119898
ℎ
119860⊺
120591119894119881
⊺
119894119898120591119898(119905) 119860
⊺
120591119894119872
⊺
119894119898(ℎ minus 120591
119898(119905)) 119860
⊺
120591119894119873
⊺
1198941198980
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minus120591119898(119905)119872
⊺
119894119898minus (ℎ minus 120591
119898(119905))119873
⊺
119894119898119881119894119898119863
119894
lowast lowast lowast lowast minus120591119898(119905) 119883
1198941198980 120591
119898(119905)119872
119894119898119863
119894
lowast lowast lowast lowast lowast minus (ℎ minus 120591119898(119905))119883
119894119898(ℎ minus 120591
119898(119905))119873
119894119898119863
119894
lowast lowast lowast lowast lowast lowast minus120575119867
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
(39)
The LMIs (20) and (21) lead to 120591119898(119905) rarr ℎ and 120591
119898(119905) rarr
0 respectively It is easy to see that Ξ1119894119898
results from Ξ119894119898|
120591119898(119905) = ℎ and Ξ
119894119898|120591119898(119905) = 0 Thus we obtain
E pound119881 (120578119905 119903
119905 119904
119905) le 120578E [119881 (120578
119905 119903
119905 119904
119905)] + 120575120596
⊺
(119905)119867120596 (119905)
(40)
Multiplying the aforementioned inequality by 119890minus120578119905 we canget
E pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] le 119890
minus120578119905
120575120596⊺
(119905)119867120596 (119905) (41)
By integrating the aforementioned inequality between 0and 119905 it follows that
119890minus120578119905
E [119881 (120578119905 119903
119905 119904
119905)] minus E [119881 (120578
0 119903
0 119904
0)]
le 120575int
119905
0
119890minus120578119904
120596⊺
(119904)119867120596 (119904) 119889119904
(42)
Denote that 119894119898
= 119877minus(12)
119875119894119898119877minus(12) 119876
119904119894=
119877minus(12)
119876119904119894119877minus(12)
(119904 = 1 2) 119876 = 119877minus(12)
119876119877minus(12)
119883119894119898
= 119877minus(12)
119883119894119898119877minus(12) 119883 = 119877
minus(12)
119883119877minus(12)
119894119898
= 119877minus(12)
119884119894119898119877minus(12) and = 119877
minus(12)
119884119877minus(12) it yields
that
E [119881 (1205780 119903
0 119904
0)]
le max119894isinN119898isinM
120582max (119894119898) 120578⊺
(0) 119877120578 (0)
+ (max119894isinN
120582max (1198761119894) +max
119894isinN120582max (1198762119894
)) 119890120575ℎ
times int
0
minusℎ
119890120575119904
120578⊺
(119904) 119877120578 (119904) 119889119904
+ 119890120575ℎ
120582max (119876)int0
minusℎ
int
0
120579
119890120575119904
120578⊺
(119904) 119877120578 (119904) 119889119904
+ 119890120575ℎ max
119894isinN119898isinM120582max (119883119894119898
)
times int
0
minusℎ
int
0
120579
119890minus120575120579
120578⊺
(119904) 119877120578 (119904) 119889119904 119889120579
+ 119890120575ℎ
120582max (119883)
times int
0
minusℎ
int
0
120579
int
0
120592
119890minus120575120579
120578⊺
(119904) 119877120578 (119904) 119889119904 119889120579 119889120592
+ 119890120575ℎ max
119894isinN119898isinM120582max (119894119898)
8 Journal of Applied Mathematics
times int
0
minusℎ
int
0
120579
119890minus120575120579
120578⊺
(119904) 119877 120578 (119904) 119889119904 119889120579
+ 119890120575ℎ
120582max ()
times int
0
minusℎ
int
0
120579
int
0
120592
119890minus120575120579
120578⊺
(119904) 119877 120578 (119904) 119889119904 119889120579 119889120592
le max119894isinN119898isinM
120582max (119894119898)
+ ℎ119890120575ℎ
(max119894isinN
120582max (1198761119894+
119894isinNmax120582max (1198762119894
)))
+ ℎ2
119890120575ℎ
( max119894isinN119898isinM
120582max (119883119894119898) + 120582max (119876)
+ max119894isinN119898isinM
120582max (119894119898))
+
1
2
ℎ3
119890120575ℎ
(120582max (119883) + 120582max ())
times supminusℎle119904le0
120578⊺
(119904) 119877120578 (119904) 120578⊺
(119904) 119877 120578 (119904)
= 1198881Λ
(43)
For given 120578 gt 0 and 0 le 119905 le 119879 we have
E [119881 (120578119905 119903
119905 119904
119905)] le E [119890
120578119905
119881 (1205780 119903
0 119904
0)]
+ 119890120578119905
120575int
119905
0
119890minus120578119904
120596⊺
(119904)119867120596 (119904) 119889119904
le 119890120578119879
1198881Λ + 119889120575119890
120578119879
120582max (119867)int119879
0
119890minus120578119904
119889119904
le 119890120578119879
1198881Λ + 119889120575120582
10
1
120578
(1 minus 119890minus120578119879
)
(44)
On the other hand it follows from (25) that
E [119881 (120578119905 119903
119905 119904
119905)] ge E [120578
⊺
(119905) 119890120582119905
119875119894119898120578 (119905)]
ge max119894isinN
120582min (119875119894119898)E [120578⊺
(119905) 119877120578 (119905)]
= 1205821E [120578
⊺
(119905) 119877120578 (119905)]
(45)
It can be derived from (43)ndash(45) that
E [120578⊺
(119905) 119877120578 (119905)] le
119890120578119879
1205821
1198881Λ + 119889120575120582
10
1
120578
(1 minus 119890minus120578119879
) (46)
From (22) and (46) we have
E [120578⊺
(119905) 119877120578 (119905)] lt 1198882 (47)
Then the system is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879)
Remark 8 In this paper 120591119898(119905) and 120591
119898(119905) may have different
upper bounds in various delay intervals satisfying (7) respec-tively However in previous work such as [20 21] 120591
119898(119905) and
120591119898(119905) are enlarged to 120591
119898(119905) le ℎ = maxℎ
119894 119898 isin M and
120591119898(119905) le 120583 = max120583
119898 119898 isin M respectively which may lead
to conservativeness inevitably However the previouse casecan be taken fully into account by employing the Lyapunov-Krasovskii functional (25)
Remark 9 When dealing with termminusint
119905
119905minusℎ
120578⊺
(119904)119884119894119898120578(119904)119889119904 the
convex combination is not employed Lemma 5 is used in thispaper and then the free-weighting matrices-dependent nulladd items are necessary to be introduced in our proof whichlead to the decrease of the number of LMIs and LMIs scalardecision variables
Remark 10 The feature of this paper is the way to dealwith the integral term Many researchers have enlarged thederivative of the Lyapunov functional in order to deal with theintegral term in mathematical operations In this paper wepropose a novel delay-dependent sufficient criterion whichensures that the Markovian jump system with different modesystems is finite-time stable
Remark 11 It should be pointed out that the novelty of theLyapunov functional (25) lies in distinct Lyapunov matrices(119875
119894119898 119883
119894119898 119884
119894119898) which is chosen for different system modes
119894 (119894 = 1 2 119873) and119898 (119898 = 1 2 119872)
Theorem 12 System (9) is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879) if there exist matrices 119875
119894gt 0 119876
119897119894gt 0 (119897 = 1 2)
119876 gt 0 119883119894119898gt 0 119883 gt 0 119884
119894119898gt 0 119884 gt 0 119878
119894119898 scalars 119888
1lt 119888
2
119879 gt 0 120582119904gt 0 (119904 = 1 2 9) 120578 gt 0 120574 gt 0 and Λ gt 0 such
that for all 119894 119895 isinN and119898 119899 isinM and (16)ndash(19) the followinginequalities hold
Σ1119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119872⊺
119894119898119875119894119898119863119894
119862
⊺
119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119872⊺
1198941198980 119864
⊺
119894
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119872⊺
119894119898119881119894119898119863119894
0
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119872119894119898119863119894
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(48)
Σ2119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119873⊺
119894119898119875119894119898119863119894
119862
⊺
119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119873⊺
1198941198980 119864
⊺
119894
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119873⊺
119894119898119881119894119898119863119894
0
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119873119894119898119863119894
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(49)
1198881Λ + 119889120575120574
21
120578
(1 minus 119890minus120578119879
) lt 1205821119890minus120578119879
1198882 (50)
Journal of Applied Mathematics 9
Proof We now consider the 119867infin
performance of system (9)Select the same Lyapunov-Krasovskii functional as Theo-rem 7 it yields that
pound119881 (120578119905 119903
119905 119904
119905) + 119890
⊺
(119905) 119890 (119905) minus 1205742
120596⊺
(119905) 120596 (119905) le 120585⊺
(119905) Σ119897119894119898120585 (119905)
(119897 = 1 2)
(51)
It follows from (49)-(50) that
E pound119881 (120578119905 119903
119905 119904
119905)
le E [120578119881 (120578119905 119903
119905 119904
119905)] + 120574
2
120596⊺
(119905) 120596 (119905) minus E [119890⊺
(119905) 119890 (119905)]
(52)
Multiplying the aforementioned inequality by 119890minus120578119905 onehas
E pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] le 119890
minus120578119905
[1205742
120596⊺
(119905) 120596 (119905) minus 119890⊺
(119905) 119890 (119905)]
(53)
In zero initial condition and E[119881(120578119905 119903
119905 119904
119905)] gt 0 by inte-
grating the aforementioned inequality between 0 and 119879 wecan get
int
119879
0
119890minus120578119905
[1205742
120596⊺
(119905) 120596 (119905) minus 119890⊺
(119905) 119890 (119905)] 119889119905
le Eint119879
0
pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] 119889119905 le 119881 (120578
0 119903
0 119904
0) = 0
(54)
Using Dynkins formula it results that
E [int119879
0
119890minus120578120592
119890⊺
(120592) 119890 (120592) 119889120592] le 1205742
E [int119879
0
119890minus120578120592
120596⊺
(120592) 120596 (120592) 119889120592]
(55)
Then it yields that
E [int119879
0
119890⊺
(120592) 119890 (120592) 119889120592] le 1205742
119890120578119879
E [int119879
0
120596⊺
(120592) 120596 (120592) 119889120592] (56)
Thus it is concluded by Definition 3 that system (9) is fi-nite-time bounded with an 119867
infinperformance 120574 This com-
pletes the proof
Remark 13 From the proof process of Theorems 7 and 12it is easy to see that neither bounding technique for cross-terms nor model transformation is involved In other wordsthe obtained result is expected to be less conservative
Remark 14 The Lyapunov asymptotic stability and finite-time stability of a class of system are independent conceptsA Lyapunov asymptotic stability system may not be finite-time stability Moreover finite-time stability system may alsonot be Lyapunov asymptotic stabilityThere exist some resultson Lyapunov stability while finite-time stability also needsour full investigation which was neglected by most previouswork
4 Finite-Time 119867infin
Filtering
Theorem 15 System (9) is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879) if there exist matrices 119875
119894119898gt 0119872
119894119898 119873
119894119898 119881
119894119898
119876119897119894gt 0 (119897 = 1 2) 119876 gt 0 119883
119894119898gt 0 119883 gt 0 119884
119894119898gt 0 119884 gt 0
119860119891119894119898
119861119891119894119898
119862119891119894119898
119872119860119894119898
119872119861119894119898
119873119860119894119898
119873119861119894119898
119881119860119894119898
119881119861119894119898
119878119894119898
scalars 1198881lt 119888
2 119879 gt 0 120590
119904gt 0 (119904 = 1 2 9) 120575 gt 0 120578 gt 0
and Λ gt 0 such that for all 119894 119895 isin N and 119898 119899 isin M thefollowing inequalities hold
119875119894119898= [
1198751119894119898
1198752119894119898
1198752119894119898
1198752119894119898
] gt 0 119872119894119898= [
1198721119894119898
1198722119894119898
1198722119894119898
1198722119894119898
] (57)
119873119894119898= [
1198731119894119898
1198732119894119898
1198732119894119898
1198732119894119898
] 119881119894119898= [
1198811119894119898
1198812119894119898
1198812119894119898
1198812119894119898
] (58)
Γ1119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Γ11119894119898
Γ12119894119898
minus119878119894119898
Γ14119894119898
Γ15119894119898
Γ16119894119898
Γ17119894119898
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
Γ24119894119898
Γ25119894119898
0 [119862⊺
119911120591119894
0]
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119872⊺
119894119898Γ46119894119898
0
lowast lowast lowast lowast minusℎ119883119894119898
Γ56119894119898
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(59)
Γ2119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Γ11119894119898
Γ12119894119898
minus119878119894119898
Γ14119894119898
Γ1015840
15119894119898Γ16119894119898
Γ17119894119898
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
Γ24119894119898
Γ1015840
251198941198980 [
119862⊺
119911120591119894
0]
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119873⊺
119894119898Γ1015840
461198941198980
lowast lowast lowast lowast minusℎ119883119894119898
Γ1015840
561198941198980
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(60)where
Γ11119894119898
= sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898+ 119890
120575ℎ
1198761119894+ 119890
120575ℎ
1198762119894+ ℎ119876 +
119890120575ℎ
minus 1
120575
119883119894119898+
119890120575ℎ
minus 120575ℎ119890120575ℎ
minus 1
1205752
119883
+
119884119894119898
ℎ
+ [
1205751198751119894119898
+ 1198751119894119898119860
119894+ 119860
⊺
1198941198751119894119898
+ 119861119891119894119898119862119910119894+ 119862
⊺
119910119894119861
⊺
119891119894119898120575119875
2119894119898+ 119860
119891119894119898+ 119860
⊺
119891119894119898
1205751198752119894119898
+ 1198752119894119898119860
119894+ 119860
⊺
1198941198752119894119898
+ 119861119891119894119898119862119910119894+ 119862
⊺
119910119894119861
⊺
119891119894119898120575119875
2119894119898+ 119860
119891119894119898+ 119860
⊺
119891119894119898
]
(61)
10 Journal of Applied Mathematics
Γ12119894119898
= minus
119884119894119898
ℎ
+ 119878119894119898+ [
1198751119894119898119860
120591119894+ 119861
119891119894119898119862119910120591119894
1198752119894119898119860
120591119894+ 119861
119891119894119898119862119910120591119894
]
Γ14119894119898
= [
119860⊺
119894119881
⊺
1119894119898+ 119862
⊺
119910119894119881
⊺
119861119894119898119881
⊺
119860119894119898
119860⊺
119894119881
⊺
2119894119898+ 119862
⊺
119910119894119881
⊺
119861119894119898119881
⊺
119860119894119898
]
Γ15119894119898
= [
ℎ119860⊺
119894119872
⊺
1119894119898+ ℎ119862
⊺
119910119894119872
⊺
119861119894119898ℎ119872
⊺
119860119894119898
ℎ119860⊺
119894119872
⊺
2119894119898+ ℎ119862
⊺
119910119894119872
⊺
119861119894119898ℎ119872
⊺
119860119894119898
]
Γ1015840
15119894119898= [
ℎ119860⊺
119894119873
⊺
1119894119898+ ℎ119862
⊺
119910119894119873
⊺
119861119894119898ℎ119873
⊺
119860119894119898
ℎ119860⊺
119894119873
⊺
2119894119898+ ℎ119862
⊺
119910119894119873
⊺
119861119894119898ℎ119873
⊺
119860119894119898
]
Γ16119894119898
= [
1198751119894119898119863
119894+ 119861
119891119894119898119863
119910119894
1198752119894119898119863
119894+ 119861
119891119894119898119863
119910119894
]
Γ17119894119898
= [
119862⊺
119911119894
minus119862119891119894119898
]
Γ24119894119898
= [
119860⊺
120591119894119881
⊺
1119894119898+ 119862
⊺
119910120591119894119881
⊺
119861119894119898
119860⊺
120591119894119881
⊺
2119894119898+ 119862
⊺
119910120591119894119881
⊺
119861119894119898
]
Γ25119894119898
= [
ℎ119860⊺
120591119894119872
⊺
1119894119898+ ℎ119862
⊺
119910120591119894119872
⊺
119861119894119898
ℎ119860⊺
120591119894119872
⊺
2119894119898+ ℎ119862
⊺
119910120591119894119872
⊺
119861119894119898
]
Γ1015840
25119894119898= [
ℎ119860⊺
120591119894119873
⊺
1119894119898+ ℎ119862
⊺
119910120591119894119873
⊺
119861119894119898
ℎ119860⊺
120591119894119873
⊺
2119894119898+ ℎ119862
⊺
119910120591119894119873
⊺
119861119894119898
]
Γ46119894119898
= [
1198811119894119898119863
119894+ 119881
119861119894119898119863
119910119894
1198812119894119898119863
119894+ 119881
119861119894119898119863
119910119894
]
Γ56119894119898
= [
ℎ1198721119894119898119863
119894+ ℎ119872
119861119894119898119863
119910119894
ℎ1198722119894119898119863
119894+ ℎ119872
119861119894119898119863
119910119894
]
Γ1015840
56119894119898= [
ℎ1198731119894119898119863
119894+ ℎ119873
119861119894119898119863
119910119894
ℎ1198732119894119898119863
119894+ ℎ119873
119861119894119898119863
119910119894
]
(62)
Then a desired filter can be chosen with parameters as
119860119891119894119898
= 119875minus1
2119894119898119860
119891119894119898 119861
119891119894119898= 119875
minus1
2119894119898119861119891119894119898
119862119891119894119898
= 119862119891119894119898
(63)
Proof We denote that
119875119894119898= [
1198751119894119898
1198752119894119898
1198752119894119898
1198752119894119898
] (64)
The term 119875119894119898119860
119894can be rewritten as
119875119894119898119860
119894= [
1198751119894119898119860
119894+ 119875
21198941198981198611198911198941198981198621199101198941198752119894119898119860
119891119894119898
1198752119894119898119860
119894+ 119875
21198941198981198611198911198941198981198621199101198941198752119894119898119860
119891119894119898
] (65)
Similarly we have
119875119894119898119861119894= [
1198751119894119898119860
120591119894+ 119875
2119894119898119861119891119894119898119862119910120591119894
1198752119894119898119860
120591119894+ 119875
2119894119898119861119891119894119898119862119910120591119894
]
119875119894119898119863
119894= [
1198751119894119898119863
119894+ 119875
2119894119898119861119891119894119898119863
119910119894
1198752119894119898119863
119894+ 119875
2119894119898119861119891119894119898119863
119910119894
]
(66)
Define 119860119891119894119898
= 1198752119894119898119860
119891119894119898 119861
119891119894119898= 119875
2119894119898119861119891119894119898
119862119891119894119898
= 119862119891119894119898
119872
119860119894119898= 119872
2119894119898119860
119891119894119898119872
119861119894119898= 119872
2119894119898119861119891119894119898
119873119860119894119898
= 1198732119894119898119860
119891119894119898
119873119861119894119898
= 1198732119894119898119861119891119894119898
119881119860119894119898
= 1198812119894119898119860
119891119894119898 and 119881
119861119894119898= 119881
2119894119898119861119891119894119898
Therefore if (59) and (60) hold system (9) is finite-timebounded with a prescribed 119867
infinperformance index 120574 The
proof is completed
Remark 16 In many actual applications the minimum valueof 1205742min is of interest In Theorem 12 with a fixed 120582 120574min canbe obtained through the following optimization procedure
min 1205742
st (48)ndash(50) (67)
In Theorem 15 as for finite-time stability and bounded-ness once the state bound 119888
2is not ascertained theminimum
value 1198882min is of interest With a fixed 120582 define 120582
1= 1 then
the following optimization problem can be formulated to getminimum value 119888
2min
min 1205891205742
+ (1 minus 120589) 1198882
st (59)-(60) and (50)
(68)
where 120589 is weighted factor and 120589 isin [0 1]
5 Illustrative Example
Example 17 Consider that the Markovian jump system andthe delay mode switching are governed by a Markov processwith the following transition rates
Λ = [
minus07 07
09 minus09] Π = [
minus1 1
12 minus12] (69)
as well as with the following parameters
1198601= [
minus35 086
minus064 minus325] 119860
2= [
minus25 034
14 minus002]
1198601205911= [
minus08 minus13
minus07 minus22] 119860
1205912= [
minus28 05
minus08 minus14]
1198631= 119863
2= [
01
02] 119862
1199101= [04 07]
1198621199102= [05 08] 119862
1199101205911= [01 02]
1198621199101205912
= [01 01]
1198631199101= 119863
1199102= 01 119862
1199111= [01 005]
1198621199112= [02 009] 119862
1199111205911= [03 06]
1198621199111205912
= [04 08] 1198631199111= 119863
1199112= 005
(70)
Journal of Applied Mathematics 11
Then we choose 119877 = 119868 119879 = 2 1198881= 1 and 119889 = 001
throughTheorem 15 it yields that themode-dependent filtersare as follows
11986011989111
= [
minus61254 15565
minus02347 minus03763]
11986011989112
= [
minus62682 14338
minus03552 minus05327]
11986011989121
= [
minus89214 24223
minus15476 minus06234]
11986011989122
= [
minus84638 28237
minus13545 minus04322]
11986111989111
= [
51465 minus28516
minus91216 95171]
11986111989112
= [
54203 minus23121
minus93156 96332]
11986111989121
= [
46193 minus11984
minus162397 163006]
11986111989122
= [
46512 minus10311
minus161132 162312]
11986211989111
= [00294 minus00397]
11986211989112
= [00271 minus003368]
11986211989121
= [01163 minus01432]
11986211989122
= [01342 minus03672]
(71)
This paper deals with the finite-time filter design problemfor a class of Markovian jump systems particularly two dif-ferentMarkov processes are considered formodeling the ran-domness of system matrix and the state delay Then throughthe numerical example we can see that results in this paperare feasible which further verified the correctness of our the-ory Therefore the paper shorten this gap
6 Conclusions
In this paper we have examined the problems of finite-time119867
infinfiltering for a class of Markovian jump systems with dif-
ferent system modes Based on a novel approach a sufficientcondition is derived such that the closed-loop Markovianjump system is finite-time bounded and satisfies a prescribedlevel of119867
infindisturbance attenuation in a finite time interval
Finally a numerical example is also given to illustrate theeffectiveness of the proposed design approach It should benoted that one of future research topics would be to investi-gate the problems of fault detection and fault tolerant controlfor time-varying Markovian jump systems with incompleteinformation over a finite-time horizon
Acknowledgments
The authors would like to thank the associate editor and theanonymous reviewers for their detailed comments and sug-gestions This work was supported by the Fund of SichuanProvincial Key Laboratory of Signal and Information Pro-cessing Xihua University (SZJJ2009-002 and SGXZD0101-10-1) and National Basic Research Program of China(2010CB732501)
References
[1] N N Krasovskiı and E A Lidskiı ldquoAnalytical design ofcontrollers in systems with random attributesmdashI Statementof the problem method of solvingrdquo Automation and RemoteControl vol 22 pp 1021ndash1025 1961
[2] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[3] L Wu X Su and P Shi ldquoSliding mode control with bounded1198712gain performance of Markovian jump singular time-delay
systemsrdquo Automatica vol 48 no 8 pp 1929ndash1933 2012[4] H Gao Z Fei J Lam and B Du ldquoFurther results on expo-
nential estimates of Markovian jump systems with mode-de-pendent time-varying delaysrdquo IEEE Transactions on AutomaticControl vol 56 no 1 pp 223ndash229 2011
[5] Z Wu H Su and J Chu ldquo119867infinfiltering for singular Markovian
jump systems with time delayrdquo International Journal of Robustand Nonlinear Control vol 20 no 8 pp 939ndash957 2010
[6] P Tino M Cernansky and L Benuskova ldquoMarkovian architec-tural bias of recurrent neural networksrdquo IEEE Transactions onNeural Networks vol 15 no 1 pp 6ndash15 2004
[7] H Dong Z Wang D W C Ho and H Gao ldquoRobust 119867infin
filtering for Markovian jump systems with randomly occurringnonlinearities and sensor saturation the finite-horizon caserdquoIEEE Transactions on Signal Processing vol 59 no 7 pp 3048ndash3057 2011
[8] Q Ma S Xu and Y Zou ldquoStability and synchronization forMarkovian jump neural networks with partly unknown tran-sition probabilitiesrdquo Neurocomputing vol 74 pp 3404ndash33412011
[9] H Zhao S Xu and Y Zou ldquoRobust119867infinfiltering for uncertain
Markovian jump systems with mode-dependent distributeddelaysrdquo International Journal of Adaptive Control and SignalProcessing vol 24 no 1 pp 83ndash94 2010
[10] Y Zhang S Xu and B Zhang ldquoDiscrete-time fuzzy Markovianjump systems with time-varying delaysrdquo IEEE Transactions onFuzzy Systems vol 17 no 2 pp 411ndash420 2009
[11] Z Wu H Su and J Chu ldquoState estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 no 10ndash12 pp 2247ndash2254 2010
[12] Q Zhu and J Cao ldquoRobust exponential stability of markovianjump impulsive stochastic Cohen-Grossberg neural networkswithmixed time delaysrdquo IEEE Transactions onNeural Networksvol 21 no 8 pp 1314ndash1325 2010
[13] H Huang G Feng and X Chen ldquoStability and stabilizationof Markovian jump systems with time delay via new Lyapunovfunctionalsrdquo IEEE Transactions on Circuits and Systems I vol59 no 10 pp 2413ndash2421 2012
12 Journal of Applied Mathematics
[14] Q Zhu and J Cao ldquoStability of Markovian jump neural net-works with impulse control and time varying delaysrdquoNonlinearAnalysis Real World Applications vol 13 no 5 pp 2259ndash22702012
[15] M Sun J Lam S Xu andY Zou ldquoRobust exponential stabiliza-tion for Markovian jump systems with mode-dependent inputdelayrdquo Automatica vol 43 no 10 pp 1799ndash1807 2007
[16] M Mohammadian A Abolmasoumi and H Momeni ldquo119867infin
mode-independent filter design for Markovian jump geneticregulatory networks with time-varying delaysrdquo Neurocomput-ing vol 87 pp 10ndash18 2012
[17] S Boyd L El Ghaoui E Feron and V Balakrishnan LinearMatrix Inequalities in SystemandControlTheory vol 15 of SIAMStudies in Applied Mathematics SIAM Philadelphia Pa USA1994
[18] S Mou H GaoW Qiang and K Chen ldquoNew delay-dependentexponential stability for neural networks with time delayrdquo IEEETransactions on Systems Man and Cybernetics Part B vol 38no 2 pp 571ndash576 2008
[19] O M Kwon J Park S Lee and E Cha ldquoAnalysis on delay-dependent stability for neural networks with time-varyingdelaysrdquo Neurocomputing vol 103 pp 114ndash120 2013
[20] L Hu H Gao andW X Zheng ldquoNovel stability of cellular neu-ral networkswith interval time-varying delayrdquoNeuralNetworksvol 21 no 10 pp 1458ndash1463 2008
[21] H Zhang Z Liu G B Huang and ZWang ldquoNovel weighting-delay-based stability criteria for recurrent neural networks withtime-varying delayrdquo IEEE Transactions on Neural Networks vol21 no 1 pp 91ndash106 2010
[22] Z Zuo C Yang and YWang ldquoA newmethod for stability anal-ysis of recurrent neural networks with interval time-varyingdelayrdquo IEEE Transactions on Neural Networks vol 21 no 2 pp339ndash344 2010
[23] P Park JWKo andC Jeong ldquoReciprocally convex approach tostability of systems with time-varying delaysrdquo Automatica vol47 no 1 pp 235ndash238 2011
[24] D Zhang L Yu and W Zhang ldquoDelay-dependent fault detec-tion for switched linear systems with time-varying delays-theaverage dwell time approachrdquo Signal Processing vol 91 pp 832ndash840 2011
[25] Z Fei H Gao and P Shi ldquoNew results on stabilization ofMark-ovian jump systems with time delayrdquoAutomatica vol 45 no 10pp 2300ndash2306 2009
[26] H R Karimi ldquoRobust delay-dependent 119867infin
control of uncer-tain time-delay systems with mixed neutral discrete anddistributed time-delays and Markovian switching parametersrdquoIEEE Transactions on Circuits and Systems I vol 58 no 8 pp1910ndash1923 2011
[27] M Luo G Liu and S Zhong ldquoRobust fault detection ofMarkovian jump systems with different systemmodesrdquoAppliedMathematical Modelling vol 37 no 7 pp 5001ndash5012 2013
[28] B Zhang andY Li ldquoExponential1198712minus119871
infinfiltering for distributed
delay systems with Markovian jumping parametersrdquo SignalProcessing vol 93 no 1 pp 206ndash216 2013
[29] A Abolmasoumi and H Momeni ldquoRobust observer-based119867infin
control of a Markovian jump system with different delay andsystem modesrdquo International Journal of Control Automationand Systems vol 9 no 4 pp 768ndash776 2011
[30] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[31] S Ma and E K Boukas ldquoRobust 119867infin
filtering for uncertaindiscrete Markov jump singular systems with mode-dependenttime delayrdquo IET Control Theory amp Applications vol 3 no 3 pp351ndash361 2009
[32] Y ChenW Bi andW Li ldquoStability analysis for neural networkswith time-varying delay a more general delay decompositionapproachrdquo Neurocomputing vol 73 no 4ndash6 pp 853ndash857 2010
[33] Z Shu J Lam and S Xu ldquoRobust stabilization of Markoviandelay systems with delay-dependent exponential estimatesrdquoAutomatica vol 42 no 11 pp 2001ndash2008 2006
[34] X Huang W Lin and B Yang ldquoGlobal finite-time stabilizationof a class of uncertain nonlinear systemsrdquo Automatica vol 41no 5 pp 881ndash888 2005
[35] C Qian and J Li ldquoGlobal finite-time stabilization by outputfeedback for planar systems without observable linearizationrdquoIEEE Transactions on Automatic Control vol 50 no 6 pp 885ndash890 2005
[36] F Amato R Ambrosino C Cosentino and G De TommasildquoInput-output finite time stabilization of linear systemsrdquo Auto-matica vol 46 no 9 pp 1558ndash1562 2010
[37] Y Hong ldquoFinite-time stabilization and stabilizability of a classof controllable systemsrdquo Systems amp Control Letters vol 46 no4 pp 231ndash236 2002
[38] F Amato and M Ariola ldquoFinite-time control of discrete-timelinear systemsrdquo IEEETransactions onAutomatic Control vol 50no 5 pp 724ndash729 2005
[39] Z Zuo Y Liu Y Wang and H Li ldquoFinite-time stochasticstability and stabilisation of linear discrete-time Markovianjump systems with partly unknown transition probabilitiesrdquoIET Control Theory amp Applications vol 6 no 10 pp 1522ndash15262012
[40] X Luan F Liu and P Shi ldquoFinite-time filtering for non-linearstochastic systems with partially known transition jump ratesrdquoIET Control Theory amp Applications vol 4 no 5 pp 735ndash7452010
[41] F Amato M Ariola and C Cosentino ldquoFinite-time control ofdiscrete-time linear systems analysis and design conditionsrdquoAutomatica vol 46 no 5 pp 919ndash924 2010
[42] X Lin H Du and S Li ldquoFinite-time boundedness and 1198712-
gain analysis for switched delay systems with norm-boundeddisturbancerdquo Applied Mathematics and Computation vol 217no 12 pp 5982ndash5993 2011
[43] J Lin S Fei and Z Gao ldquoStabilization of discrete-time switchedsingular time-delay systems under asynchronous switchingrdquoJournal of the Franklin Institute vol 349 no 5 pp 1808ndash18272012
[44] H Song L Yu D Zhang and W-A Zhang ldquoFinite-time 119867infin
control for a class of discrete-time switched time-delay systemswith quantized feedbackrdquoCommunications in Nonlinear Scienceand Numerical Simulation vol 17 no 12 pp 4802ndash4814 2012
[45] Z Zuo H Li Y Liu and Y Wang ldquoOn finite-time stochasticstability and stabilization ofMarkovian jump systems subject topartial information on transition probabilitiesrdquoCircuits Systemsand Signal Processing 2012
[46] H Liu Y Shen and X Zhao ldquoDelay-dependent observer-based119867
infinfinite-time control for switched systems with time-varying
delayrdquoNonlinear Analysis Hybrid Systems vol 6 no 3 pp 885ndash898 2012
[47] Y Zhang C Liu and X Mu ldquoRobust finite-time119867infincontrol of
singular stochastic systems via static output feedbackrdquo AppliedMathematics and Computation vol 218 no 9 pp 5629ndash56402012
Journal of Applied Mathematics 13
[48] J Cheng H Zhu S Zhong Y Zhang and Y Zeng ldquoFinite-timeStabilization of 119867
infinfiltering for switched stochastic systemsrdquo
Circuits Systems and Signal Processing 2012[49] J Cheng H Zhu S Zhong and Y Zhang ldquoFinite-time bound-
ness of 119867infin
filtering for switching discrete-time systemsrdquo In-ternational Journal of Control Automation and Systems vol 10pp 1129ndash1135 2012
[50] Z Xiang C Qiao and M S Mahmoud ldquoFinite-time analysisand119867
infincontrol for switched stochastic systemsrdquo Journal of the
Franklin Institute vol 349 no 3 pp 915ndash927 2012
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
4 Journal of Applied Mathematics
sum
119899isinM119898 = 119899
120587119898119899119884119894119899+ sum
119895isinN119894 = 119895
120582119894119895119884119895119898minus 119884 lt 0 (18)
[
[
[
119884119894119898
ℎ
119878119894119898
lowast
119884119894119898
ℎ
]
]
]
gt 0 (19)
Ξ1119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119872⊺
119894119898119875119894119898119863119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119872⊺
1198941198980
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119872⊺
119894119898119881119894119898119863119894
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119872119894119898119863119894
lowast lowast lowast lowast lowast minus120575119867
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(20)
Ξ2119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119872⊺
119894119898119875119894119898119863119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119873⊺
1198941198980
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119873⊺
119894119898119881119894119898119863119894
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119873119894119898119863119894
lowast lowast lowast lowast lowast minus120575119867
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(21)
1198881Λ + 119889120575120582
10
1
120578
(1 minus 119890minus120578119879
) lt 1205821119890minus120578119879
1198882 (22)
where
Ξ11119894119898
= sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898
+ 120575119875119894119898+ 119875
119894119898119860
119894+ 119860
⊺
119894119875119894119898+ 119890
120575ℎ
1198761119894
+ 119890120575ℎ
1198762119894+ ℎ119876
119890120575ℎ
minus 1
120575
119883119894119898
+
119890120575ℎ
minus 120575ℎ119890120575ℎ
minus 1
1205752
119883 +
119884119894119898
ℎ
Ξ12119894119898
= 119875119894119898119861119894minus
119884119894119898
ℎ
+ 119878119894119898
Ξ22119894119898
= 119903 (120583119898) 119876
2119894+
2119884119894119898
ℎ
minus 119878119894119898minus 119878
⊺
119894119898
Ξ44119894119898
= minus119881119894119898minus 119881
⊺
119894119898+
119890120575ℎ
minus 1
120575
119884119894119898+
119890120575ℎ
minus 120575ℎ119890120575ℎ
minus 1
1205752
119884
119903 (120591119898) =
minus (1 minus 120583119898) 119890
120575ℎ119894 if 120583
119898gt 1
minus (1 minus 120583119898) if 120583
119898le 1
Λ = 1205822+ ℎ119890
120575ℎ
(1205823+ 120582
4) + ℎ
2
119890120575ℎ
(1205825+ 120582
6+ 120582
8)
+
1
2
ℎ3
119890120575ℎ
(1205827+ 120582
9)
1205821= min
119894isinN119898isinM120582min (119894119898) 120582
2= max
119894isinN119898isinM120582max (119894119898)
1205823= max
119894isinN120582max (1198761119894
)
1205824= max
119894isinN120582max (1198762119894
) 1205825= 120582max (119876)
1205826= max
119894isinN119898isinM120582max (119883119894119898
) 1205827= 120582max (119883)
1205828= max
119894isinN119898isinM120582max (119894119898) 120582
9= 120582max ()
12058210= 120582max (119867)
119894119898= 119877
minus(12)
119875119894119898119877minus(12)
119876119897119894= 119877
minus(12)
119876119897119894119877minus(12)
(119897 = 1 2)
119876 = 119877minus(12)
119876119877minus(12)
119883119894119898= 119877
minus(12)
119883119894119898119877minus(12)
119883 = 119877minus(12)
119883119877minus(12)
119894119898= 119877
minus(12)
119884119894119898119877minus(12)
= 119877minus(12)
119884119877minus(12)
(23)
Proof First in order to cast our model into the framework oftheMarkov processes we define a new process (120578
119905 119903
119905 119904
119905) 119905 ge
0 by
120578119905(119904) = 120578 (119905 + 119904) 119904 isin [minusℎ 0] (24)
Now we consider the following Lyapunov-Krasovskiifunctional
119881 (120578119905 119903
119905 119904
119905) =
4
sum
119897=1
119881119897(120578
119905 119903
119905 119905) (25)
where
1198811(120578
119905 119903
119905 119904
119905) = 120578(119905)
⊺
119890120575119905
119875119903119905119904119905
120578 (119905)
1198812(120578
119905 119903
119905 119904
119905) = int
119905
119905minusℎ
119890120575(119904+ℎ)
120578⊺
(119904) 1198761119903119905
120578 (119904) 119889119904
+ int
119905
119905minus120591119904119905(119905)
119890120575(119904+ℎ)
120578⊺
(119904) 1198762119903119905
120578 (119904) 119889119904
+ int
0
minusℎ
int
119905
119905+120579
119890120575(119904+ℎ)
120578⊺
(119904) 119876120578 (119904) 119889119904 119889120579
1198813(120578
119905 119903
119905 119904
119905) = int
0
minusℎ
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904)119883119903119905119904119905
120578 (119904) 119889119904 119889120579
+ int
0
minusℎ
int
0
120579
int
119905
119905+120592
119890120575(119904minus120579)
120578⊺
(119904)119883120578 (119904) 119889119904 119889120592 119889120579
1198814(120578
119905 119903
119905 119904
119905) = int
0
minusℎ
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904) 119884119903119905119904119905
120578 (119904) 119889119904 119889120579
+ int
0
minusℎ
int
0
120579
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904) 119884 120578 (119904) 119889119904 119889120592 119889120579
(26)
Journal of Applied Mathematics 5
Then for each 119903119905= 119894 119904
119905= 119898 we have
pound1198811(120578
119905 119894 119898)
= limΔrarr0
+
1
Δ
E 120578⊺
(119905 + Δ) 119890120575(119905+Δ)
119875119903119905+Δ
119904119905+Δ
120578 (119905 + Δ)
minus120578⊺
(119905) 119890120575119905
119875119894119898120578 (119905)
= limΔrarr0
+
1
Δ
E
120578⊺
(119905 + Δ)[
[
(1 + 120582119894119894Δ + 119900 (Δ))
times (1 + 120587119898119898Δ + 119900 (Δ)) 119890
120575(119905+Δ)
119875119894119898
+ (1 + 120582119894119894Δ + 119900 (Δ))
times ( sum
119899isinM
(120587119898119899Δ + 119900 (Δ))) 119890
120575(119905+Δ)
119875119894119899
+ (1 + 120587119898119898Δ + 119900 (Δ))
times (sum
119895isinN
(120582119894119895Δ + 119900 (Δ))) 119890
120575(119905+Δ)
119875119895119898
+ ( sum
119899isinM
(120587119898119899Δ + 119900 (Δ)))
times(sum
119895isinN
(120582119894119895Δ + 119900 (Δ))) 119890
120575(119905+Δ)
119875119895119899
]
]
times 120578 (119905 + Δ)
minus 120578⊺
(119905) 119890120575119905
119875119894119898120578 (119905)
= 120575119890120575119905
120578⊺
(119905) 119875119894119898120578 (119905) + 2119890
120575119905
120578⊺
(119905) 119875119894119898120578 (119905)
+ 119890120575119905
120578⊺
(119905)( sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898)120578 (119905)
= 119890120575119905
120578⊺
(119905)( sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898+ 120575119875
119894119898
+ 119875119894119898119860
119894+ 119860
⊺
119894119875119894119898)120578 (119905)
+ 2119890120575119905
120578⊺
(119905) 119875119894119898119861119894120578 (119905 minus 120591
119898(119905)) + 2119890
120575119905
120578⊺
(119905) 119875119894119898119863
119894120596 (119905)
(27)pound119881
2(120578
119905 119894 119898)
= limΔrarr0
+
1
Δ
Eint119905+Δ
119905+Δminusℎ
120578⊺
(119904) 119890120575(119904+ℎ)
1198761119903119905+Δ
120578 (119904) 119889119904
minusint
119905
119905minusℎ
120578⊺
(119904) 119890120575(119904+ℎ)
1198761119894120578 (119904) 119889119904
+ limΔrarr0
+
1
Δ
Eint119905+Δ
119905+Δminus120591119904119905+Δ
(119905)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119903119905+Δ
120578 (119904) 119889119904
minusint
119905
119905minus120591119898(119905)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119894120578 (119904) 119889119904
+ limΔrarr0
+
1
Δ
Eint0
minusℎ
int
119905+Δ
119905+Δ+120579
119890120575(119904+ℎ)
120578⊺
(119904) 119876120578 (119904) 119889119904 119889120579
minusint
0
minusℎ
int
119905
119905+120579
119890120575(119904+ℎ)
120578⊺
(119904) 119876120578 (119904) 119889119904 119889120579
= limΔrarr0
+
1
Δ
Eint119905+Δ
119905+Δminusℎ
120578⊺
(119904) 119890120575(119904+ℎ)
1198761119894120578 (119904) 119889119904
+ int
119905+Δ
119905+Δminusℎ
120578⊺
(119904)
times sum
119895isinN
(120582119894119895Δ + 119900 (Δ))
times 119890120575(119904+ℎ)
1198761119894120578 (119904) 119889119904
minusint
119905
119905minusℎ
120578⊺
(119904) 1198761119894120578 (119904) 119889119904
+ limΔrarr0
+
1
Δ
E[int119905+Δ
119905+Δminus120591119895(119905+Δ)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119894120578 (119904) 119889119904
+ sum
119899isinM
(120587119898119899Δ + 119900 (Δ))
times int
119905
119905+Δminus120591119899(119905+Δ)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119894120578 (119904) 119889119904
+ int
119905+Δ
minus120591119895(119905 + Δ)
119905+Δ
120578⊺
(119904)
times sum
119895isinN
(120582119894119895Δ + 119900 (Δ))
times 119890120575(119904+ℎ)
1198762119895120578 (119904) 119889119904
minusint
119905
119905minus120591119898(119905)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119894120578 (119904) 119889119904]
+ 119890120575119905
120578⊺
(119905) ℎ119876120578 (119905) minus int
119905minusℎ
119890120575(119904+ℎ)
120578⊺
(119904) 119876120578 (119904) 119889119904
le 119890120575119905
120578⊺
(119905) (119890120575ℎ
1198761119894+ 119890
120575ℎ
1198762119894+ ℎ119876) 120578 (119905)
minus 119890120575119905
120578⊺
(119905 minus ℎ)1198761119894120578 (119905 minus ℎ)
minus (1 minus 120591119898(119905)) 119890
120575119905
120578⊺
(119905 minus 120591119898(119905)) 119876
2119894120578 (119905 minus 120591
119898(119905))
+ int
119905
119905minusℎ
119890120590(119904+ℎ)
120578⊺
(119904)( sum
119895isinN119895 = 119894
120582119894119895(119876
1119895+ 119876
2119895)
6 Journal of Applied Mathematics
+ sum
119899isinM119898 = 119899
120587119898119899119876
2119894minus 119876)
times 120578 (119904) 119889119904
(28)
Since
0 le 120591119898(119905) le ℎ
119898 (29)
we define
119903 (120591119898) =
minus (1 minus 120583119898) 119890
120575ℎ119898 if 120583
119898gt 1
minus (1 minus 120583119898) if 120583
119898le 1
(30)
Then
pound1198812(120578
119905 119894 119898) le 119890
120575119905
120578⊺
(119905) (119890120575ℎ
1198761119894+ 119890
120575ℎ
1198762119894+ ℎ119876) 120578 (119905)
minus 119890120575119905
120578⊺
(119905 minus ℎ)1198761119894120578 (119905 minus ℎ)
+ 119890120575119905
120578⊺
(119905 minus 120591119898(119905)) 119903 (120591
119898) 119876
2119894120578 (119905 minus 120591
119898(119905))
+ int
119905
119905minusℎ
119890120590(119904+ℎ)
120578⊺
(119904)
times ( sum
119895isinN119895 = 119894
120582119894119895(119876
1119895+ 119876
2119895)
+ sum
119899isinM119898 = 119899
120587119898119899119876
2119894minus 119876)
times 120578 (119904) 119889119904
(31)
Similar to the previous process we can obtain
pound1198813(120578
119905 119894 119898) le int
0
minusℎ
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904)
times ( sum
119899isinM119898 = 119899
120587119898119899119883
119894119899
+ sum
119895isinN119894 = 119895
120582119894119895119883
119895119898minus 119883)
times 120578 (119904) 119889119904 119889120579
+ 119890120575119905
120578⊺
(119905) 119883119894119898120578 (119905) int
0
minusℎ
119890minus120575120592
119889120592
minus 119890120575119905
int
119905
119905minusℎ
120578⊺
(119904)119883119894119898120578 (119904) 119889119904
+ 119890120575119905
120578⊺
(119905) 119883120578 (119905) int
0
minusℎ
int
0
120592
119890minus120575120592
119889120579119889120592
(32)
By using Lemma 5 it yields that
minusint
119905
119905minusℎ
120578⊺
(119904)119883119894119898120578 (119904) 119889119904 = minusint
119905
119905minus120591119898(119905)
120578⊺
(119904)119883119894119898120578 (119904) 119889119904
minus int
119905minus120591119898(119905)
119905minusℎ
120578⊺
(119904)119883119894119898120578 (119904) 119889119904
le minus120591119898(119905) 119880
⊺
1119883
1198941198981198801
minus (ℎ minus 120591119898(119905)) 119880
⊺
2119883
1198941198981198802
(33)
where
1198801=
1
120591119898(119905)
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904
1198802=
1
ℎ minus 120591119898(119905)
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904
lim120591119898(119905)rarr0
1
120591119898(119905)
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904 = 120578 (119905)
lim120591119898(119905)rarrℎ
1
ℎ minus 120591119898(119905)
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904 = 120578 (119905 minus ℎ)
(34)
From the Newton-Leibniz formula the following equa-tion is true for any matrices119872
119894119898 119873
119894119898 and 119881
119894119898with appro-
priate dimensions
(2120591119898(119905) 119880
⊺
1119872
119894119898+ 2 (ℎ minus 120591
119898(119905)) 119880
⊺
2119873
119894119898+ 2 120578
⊺
(119905) 119881119894119898)
times [minus 120578 (119905) + 119860119894120578 (119905) + 119861
120591119894120578 (119905 minus 120591
119898(119905)) + 119863
119894120596 (119905)] = 0
(35)
pound1198814(120578
119905 119894 119898)
le int
0
minusℎ
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904)
times ( sum
119899isinM119898 = 119899
120587119898119899119884119894119899
+ sum
119895isinN119894 = 119895
120582119894119895119884119895119898minus 119884)
times 120578 (119904) 119889119904 119889120579
+ 119890120575119905
120578⊺
(119905) 119884119894119898120578 (119905) int
0
minusℎ
119890minus120575120592
119889120592
minus 119890120575119905
int
119905
119905minusℎ
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
+ 119890120575119905
120578⊺
(119905) 119884 120578 (119905) int
0
minusℎ
int
0
120592
119890minus120575120592
119889120579 119889120592
(36)
Journal of Applied Mathematics 7
From Lemma 5 it yields that
minus int
119905
119905minusℎ
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
= minusint
119905
119905minus120591119898(119905)
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
minus int
119905minus120591119898(119905)
119905minusℎ
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
le minus
ℎ
120591119898(119905)
[int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904]
⊺
119884119894119898
ℎ
[int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904]
minus
ℎ
ℎ minus 120591119898(119905)
[int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904]
⊺
119884119894119898
ℎ
[int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904]
le minus
[
[
[
[
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904
]
]
]
]
⊺
[
[
[
119884119894119898
ℎ
119878119894119898
lowast
119884119894119898
ℎ
]
]
]
[
[
[
[
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904
]
]
]
]
(37)
From (25)ndash(37) we can eventually obtain
pound119881 (120578119905 119903
119905 119904
119905) minus 120575120596
⊺
(119905)119867120596 (119905) le 119890120575119905
120585⊺
(119905) Ξ119894119898120585 (119905) (38)
where
120585⊺
(119905) = [120578⊺
(119905) 120578⊺
(119905 minus 120591119898(119905)) 120578
⊺
(119905 minus ℎ) 120578⊺
(119905) 119880⊺
1 119880
⊺
2 120596
⊺
(119905)]
Ξ119894119898=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860⊺
119894119881
⊺
119894119898120591119898(119905) 119860
⊺
119894119872
⊺
119894119898(ℎ minus 120591
119898(119905)) 119860
⊺
119894119873
⊺
119894119898119875119894119898119863
119894
lowast Ξ22119894119898
119878119894119898minus
119884119894119898
ℎ
119860⊺
120591119894119881
⊺
119894119898120591119898(119905) 119860
⊺
120591119894119872
⊺
119894119898(ℎ minus 120591
119898(119905)) 119860
⊺
120591119894119873
⊺
1198941198980
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minus120591119898(119905)119872
⊺
119894119898minus (ℎ minus 120591
119898(119905))119873
⊺
119894119898119881119894119898119863
119894
lowast lowast lowast lowast minus120591119898(119905) 119883
1198941198980 120591
119898(119905)119872
119894119898119863
119894
lowast lowast lowast lowast lowast minus (ℎ minus 120591119898(119905))119883
119894119898(ℎ minus 120591
119898(119905))119873
119894119898119863
119894
lowast lowast lowast lowast lowast lowast minus120575119867
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
(39)
The LMIs (20) and (21) lead to 120591119898(119905) rarr ℎ and 120591
119898(119905) rarr
0 respectively It is easy to see that Ξ1119894119898
results from Ξ119894119898|
120591119898(119905) = ℎ and Ξ
119894119898|120591119898(119905) = 0 Thus we obtain
E pound119881 (120578119905 119903
119905 119904
119905) le 120578E [119881 (120578
119905 119903
119905 119904
119905)] + 120575120596
⊺
(119905)119867120596 (119905)
(40)
Multiplying the aforementioned inequality by 119890minus120578119905 we canget
E pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] le 119890
minus120578119905
120575120596⊺
(119905)119867120596 (119905) (41)
By integrating the aforementioned inequality between 0and 119905 it follows that
119890minus120578119905
E [119881 (120578119905 119903
119905 119904
119905)] minus E [119881 (120578
0 119903
0 119904
0)]
le 120575int
119905
0
119890minus120578119904
120596⊺
(119904)119867120596 (119904) 119889119904
(42)
Denote that 119894119898
= 119877minus(12)
119875119894119898119877minus(12) 119876
119904119894=
119877minus(12)
119876119904119894119877minus(12)
(119904 = 1 2) 119876 = 119877minus(12)
119876119877minus(12)
119883119894119898
= 119877minus(12)
119883119894119898119877minus(12) 119883 = 119877
minus(12)
119883119877minus(12)
119894119898
= 119877minus(12)
119884119894119898119877minus(12) and = 119877
minus(12)
119884119877minus(12) it yields
that
E [119881 (1205780 119903
0 119904
0)]
le max119894isinN119898isinM
120582max (119894119898) 120578⊺
(0) 119877120578 (0)
+ (max119894isinN
120582max (1198761119894) +max
119894isinN120582max (1198762119894
)) 119890120575ℎ
times int
0
minusℎ
119890120575119904
120578⊺
(119904) 119877120578 (119904) 119889119904
+ 119890120575ℎ
120582max (119876)int0
minusℎ
int
0
120579
119890120575119904
120578⊺
(119904) 119877120578 (119904) 119889119904
+ 119890120575ℎ max
119894isinN119898isinM120582max (119883119894119898
)
times int
0
minusℎ
int
0
120579
119890minus120575120579
120578⊺
(119904) 119877120578 (119904) 119889119904 119889120579
+ 119890120575ℎ
120582max (119883)
times int
0
minusℎ
int
0
120579
int
0
120592
119890minus120575120579
120578⊺
(119904) 119877120578 (119904) 119889119904 119889120579 119889120592
+ 119890120575ℎ max
119894isinN119898isinM120582max (119894119898)
8 Journal of Applied Mathematics
times int
0
minusℎ
int
0
120579
119890minus120575120579
120578⊺
(119904) 119877 120578 (119904) 119889119904 119889120579
+ 119890120575ℎ
120582max ()
times int
0
minusℎ
int
0
120579
int
0
120592
119890minus120575120579
120578⊺
(119904) 119877 120578 (119904) 119889119904 119889120579 119889120592
le max119894isinN119898isinM
120582max (119894119898)
+ ℎ119890120575ℎ
(max119894isinN
120582max (1198761119894+
119894isinNmax120582max (1198762119894
)))
+ ℎ2
119890120575ℎ
( max119894isinN119898isinM
120582max (119883119894119898) + 120582max (119876)
+ max119894isinN119898isinM
120582max (119894119898))
+
1
2
ℎ3
119890120575ℎ
(120582max (119883) + 120582max ())
times supminusℎle119904le0
120578⊺
(119904) 119877120578 (119904) 120578⊺
(119904) 119877 120578 (119904)
= 1198881Λ
(43)
For given 120578 gt 0 and 0 le 119905 le 119879 we have
E [119881 (120578119905 119903
119905 119904
119905)] le E [119890
120578119905
119881 (1205780 119903
0 119904
0)]
+ 119890120578119905
120575int
119905
0
119890minus120578119904
120596⊺
(119904)119867120596 (119904) 119889119904
le 119890120578119879
1198881Λ + 119889120575119890
120578119879
120582max (119867)int119879
0
119890minus120578119904
119889119904
le 119890120578119879
1198881Λ + 119889120575120582
10
1
120578
(1 minus 119890minus120578119879
)
(44)
On the other hand it follows from (25) that
E [119881 (120578119905 119903
119905 119904
119905)] ge E [120578
⊺
(119905) 119890120582119905
119875119894119898120578 (119905)]
ge max119894isinN
120582min (119875119894119898)E [120578⊺
(119905) 119877120578 (119905)]
= 1205821E [120578
⊺
(119905) 119877120578 (119905)]
(45)
It can be derived from (43)ndash(45) that
E [120578⊺
(119905) 119877120578 (119905)] le
119890120578119879
1205821
1198881Λ + 119889120575120582
10
1
120578
(1 minus 119890minus120578119879
) (46)
From (22) and (46) we have
E [120578⊺
(119905) 119877120578 (119905)] lt 1198882 (47)
Then the system is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879)
Remark 8 In this paper 120591119898(119905) and 120591
119898(119905) may have different
upper bounds in various delay intervals satisfying (7) respec-tively However in previous work such as [20 21] 120591
119898(119905) and
120591119898(119905) are enlarged to 120591
119898(119905) le ℎ = maxℎ
119894 119898 isin M and
120591119898(119905) le 120583 = max120583
119898 119898 isin M respectively which may lead
to conservativeness inevitably However the previouse casecan be taken fully into account by employing the Lyapunov-Krasovskii functional (25)
Remark 9 When dealing with termminusint
119905
119905minusℎ
120578⊺
(119904)119884119894119898120578(119904)119889119904 the
convex combination is not employed Lemma 5 is used in thispaper and then the free-weighting matrices-dependent nulladd items are necessary to be introduced in our proof whichlead to the decrease of the number of LMIs and LMIs scalardecision variables
Remark 10 The feature of this paper is the way to dealwith the integral term Many researchers have enlarged thederivative of the Lyapunov functional in order to deal with theintegral term in mathematical operations In this paper wepropose a novel delay-dependent sufficient criterion whichensures that the Markovian jump system with different modesystems is finite-time stable
Remark 11 It should be pointed out that the novelty of theLyapunov functional (25) lies in distinct Lyapunov matrices(119875
119894119898 119883
119894119898 119884
119894119898) which is chosen for different system modes
119894 (119894 = 1 2 119873) and119898 (119898 = 1 2 119872)
Theorem 12 System (9) is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879) if there exist matrices 119875
119894gt 0 119876
119897119894gt 0 (119897 = 1 2)
119876 gt 0 119883119894119898gt 0 119883 gt 0 119884
119894119898gt 0 119884 gt 0 119878
119894119898 scalars 119888
1lt 119888
2
119879 gt 0 120582119904gt 0 (119904 = 1 2 9) 120578 gt 0 120574 gt 0 and Λ gt 0 such
that for all 119894 119895 isinN and119898 119899 isinM and (16)ndash(19) the followinginequalities hold
Σ1119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119872⊺
119894119898119875119894119898119863119894
119862
⊺
119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119872⊺
1198941198980 119864
⊺
119894
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119872⊺
119894119898119881119894119898119863119894
0
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119872119894119898119863119894
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(48)
Σ2119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119873⊺
119894119898119875119894119898119863119894
119862
⊺
119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119873⊺
1198941198980 119864
⊺
119894
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119873⊺
119894119898119881119894119898119863119894
0
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119873119894119898119863119894
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(49)
1198881Λ + 119889120575120574
21
120578
(1 minus 119890minus120578119879
) lt 1205821119890minus120578119879
1198882 (50)
Journal of Applied Mathematics 9
Proof We now consider the 119867infin
performance of system (9)Select the same Lyapunov-Krasovskii functional as Theo-rem 7 it yields that
pound119881 (120578119905 119903
119905 119904
119905) + 119890
⊺
(119905) 119890 (119905) minus 1205742
120596⊺
(119905) 120596 (119905) le 120585⊺
(119905) Σ119897119894119898120585 (119905)
(119897 = 1 2)
(51)
It follows from (49)-(50) that
E pound119881 (120578119905 119903
119905 119904
119905)
le E [120578119881 (120578119905 119903
119905 119904
119905)] + 120574
2
120596⊺
(119905) 120596 (119905) minus E [119890⊺
(119905) 119890 (119905)]
(52)
Multiplying the aforementioned inequality by 119890minus120578119905 onehas
E pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] le 119890
minus120578119905
[1205742
120596⊺
(119905) 120596 (119905) minus 119890⊺
(119905) 119890 (119905)]
(53)
In zero initial condition and E[119881(120578119905 119903
119905 119904
119905)] gt 0 by inte-
grating the aforementioned inequality between 0 and 119879 wecan get
int
119879
0
119890minus120578119905
[1205742
120596⊺
(119905) 120596 (119905) minus 119890⊺
(119905) 119890 (119905)] 119889119905
le Eint119879
0
pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] 119889119905 le 119881 (120578
0 119903
0 119904
0) = 0
(54)
Using Dynkins formula it results that
E [int119879
0
119890minus120578120592
119890⊺
(120592) 119890 (120592) 119889120592] le 1205742
E [int119879
0
119890minus120578120592
120596⊺
(120592) 120596 (120592) 119889120592]
(55)
Then it yields that
E [int119879
0
119890⊺
(120592) 119890 (120592) 119889120592] le 1205742
119890120578119879
E [int119879
0
120596⊺
(120592) 120596 (120592) 119889120592] (56)
Thus it is concluded by Definition 3 that system (9) is fi-nite-time bounded with an 119867
infinperformance 120574 This com-
pletes the proof
Remark 13 From the proof process of Theorems 7 and 12it is easy to see that neither bounding technique for cross-terms nor model transformation is involved In other wordsthe obtained result is expected to be less conservative
Remark 14 The Lyapunov asymptotic stability and finite-time stability of a class of system are independent conceptsA Lyapunov asymptotic stability system may not be finite-time stability Moreover finite-time stability system may alsonot be Lyapunov asymptotic stabilityThere exist some resultson Lyapunov stability while finite-time stability also needsour full investigation which was neglected by most previouswork
4 Finite-Time 119867infin
Filtering
Theorem 15 System (9) is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879) if there exist matrices 119875
119894119898gt 0119872
119894119898 119873
119894119898 119881
119894119898
119876119897119894gt 0 (119897 = 1 2) 119876 gt 0 119883
119894119898gt 0 119883 gt 0 119884
119894119898gt 0 119884 gt 0
119860119891119894119898
119861119891119894119898
119862119891119894119898
119872119860119894119898
119872119861119894119898
119873119860119894119898
119873119861119894119898
119881119860119894119898
119881119861119894119898
119878119894119898
scalars 1198881lt 119888
2 119879 gt 0 120590
119904gt 0 (119904 = 1 2 9) 120575 gt 0 120578 gt 0
and Λ gt 0 such that for all 119894 119895 isin N and 119898 119899 isin M thefollowing inequalities hold
119875119894119898= [
1198751119894119898
1198752119894119898
1198752119894119898
1198752119894119898
] gt 0 119872119894119898= [
1198721119894119898
1198722119894119898
1198722119894119898
1198722119894119898
] (57)
119873119894119898= [
1198731119894119898
1198732119894119898
1198732119894119898
1198732119894119898
] 119881119894119898= [
1198811119894119898
1198812119894119898
1198812119894119898
1198812119894119898
] (58)
Γ1119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Γ11119894119898
Γ12119894119898
minus119878119894119898
Γ14119894119898
Γ15119894119898
Γ16119894119898
Γ17119894119898
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
Γ24119894119898
Γ25119894119898
0 [119862⊺
119911120591119894
0]
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119872⊺
119894119898Γ46119894119898
0
lowast lowast lowast lowast minusℎ119883119894119898
Γ56119894119898
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(59)
Γ2119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Γ11119894119898
Γ12119894119898
minus119878119894119898
Γ14119894119898
Γ1015840
15119894119898Γ16119894119898
Γ17119894119898
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
Γ24119894119898
Γ1015840
251198941198980 [
119862⊺
119911120591119894
0]
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119873⊺
119894119898Γ1015840
461198941198980
lowast lowast lowast lowast minusℎ119883119894119898
Γ1015840
561198941198980
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(60)where
Γ11119894119898
= sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898+ 119890
120575ℎ
1198761119894+ 119890
120575ℎ
1198762119894+ ℎ119876 +
119890120575ℎ
minus 1
120575
119883119894119898+
119890120575ℎ
minus 120575ℎ119890120575ℎ
minus 1
1205752
119883
+
119884119894119898
ℎ
+ [
1205751198751119894119898
+ 1198751119894119898119860
119894+ 119860
⊺
1198941198751119894119898
+ 119861119891119894119898119862119910119894+ 119862
⊺
119910119894119861
⊺
119891119894119898120575119875
2119894119898+ 119860
119891119894119898+ 119860
⊺
119891119894119898
1205751198752119894119898
+ 1198752119894119898119860
119894+ 119860
⊺
1198941198752119894119898
+ 119861119891119894119898119862119910119894+ 119862
⊺
119910119894119861
⊺
119891119894119898120575119875
2119894119898+ 119860
119891119894119898+ 119860
⊺
119891119894119898
]
(61)
10 Journal of Applied Mathematics
Γ12119894119898
= minus
119884119894119898
ℎ
+ 119878119894119898+ [
1198751119894119898119860
120591119894+ 119861
119891119894119898119862119910120591119894
1198752119894119898119860
120591119894+ 119861
119891119894119898119862119910120591119894
]
Γ14119894119898
= [
119860⊺
119894119881
⊺
1119894119898+ 119862
⊺
119910119894119881
⊺
119861119894119898119881
⊺
119860119894119898
119860⊺
119894119881
⊺
2119894119898+ 119862
⊺
119910119894119881
⊺
119861119894119898119881
⊺
119860119894119898
]
Γ15119894119898
= [
ℎ119860⊺
119894119872
⊺
1119894119898+ ℎ119862
⊺
119910119894119872
⊺
119861119894119898ℎ119872
⊺
119860119894119898
ℎ119860⊺
119894119872
⊺
2119894119898+ ℎ119862
⊺
119910119894119872
⊺
119861119894119898ℎ119872
⊺
119860119894119898
]
Γ1015840
15119894119898= [
ℎ119860⊺
119894119873
⊺
1119894119898+ ℎ119862
⊺
119910119894119873
⊺
119861119894119898ℎ119873
⊺
119860119894119898
ℎ119860⊺
119894119873
⊺
2119894119898+ ℎ119862
⊺
119910119894119873
⊺
119861119894119898ℎ119873
⊺
119860119894119898
]
Γ16119894119898
= [
1198751119894119898119863
119894+ 119861
119891119894119898119863
119910119894
1198752119894119898119863
119894+ 119861
119891119894119898119863
119910119894
]
Γ17119894119898
= [
119862⊺
119911119894
minus119862119891119894119898
]
Γ24119894119898
= [
119860⊺
120591119894119881
⊺
1119894119898+ 119862
⊺
119910120591119894119881
⊺
119861119894119898
119860⊺
120591119894119881
⊺
2119894119898+ 119862
⊺
119910120591119894119881
⊺
119861119894119898
]
Γ25119894119898
= [
ℎ119860⊺
120591119894119872
⊺
1119894119898+ ℎ119862
⊺
119910120591119894119872
⊺
119861119894119898
ℎ119860⊺
120591119894119872
⊺
2119894119898+ ℎ119862
⊺
119910120591119894119872
⊺
119861119894119898
]
Γ1015840
25119894119898= [
ℎ119860⊺
120591119894119873
⊺
1119894119898+ ℎ119862
⊺
119910120591119894119873
⊺
119861119894119898
ℎ119860⊺
120591119894119873
⊺
2119894119898+ ℎ119862
⊺
119910120591119894119873
⊺
119861119894119898
]
Γ46119894119898
= [
1198811119894119898119863
119894+ 119881
119861119894119898119863
119910119894
1198812119894119898119863
119894+ 119881
119861119894119898119863
119910119894
]
Γ56119894119898
= [
ℎ1198721119894119898119863
119894+ ℎ119872
119861119894119898119863
119910119894
ℎ1198722119894119898119863
119894+ ℎ119872
119861119894119898119863
119910119894
]
Γ1015840
56119894119898= [
ℎ1198731119894119898119863
119894+ ℎ119873
119861119894119898119863
119910119894
ℎ1198732119894119898119863
119894+ ℎ119873
119861119894119898119863
119910119894
]
(62)
Then a desired filter can be chosen with parameters as
119860119891119894119898
= 119875minus1
2119894119898119860
119891119894119898 119861
119891119894119898= 119875
minus1
2119894119898119861119891119894119898
119862119891119894119898
= 119862119891119894119898
(63)
Proof We denote that
119875119894119898= [
1198751119894119898
1198752119894119898
1198752119894119898
1198752119894119898
] (64)
The term 119875119894119898119860
119894can be rewritten as
119875119894119898119860
119894= [
1198751119894119898119860
119894+ 119875
21198941198981198611198911198941198981198621199101198941198752119894119898119860
119891119894119898
1198752119894119898119860
119894+ 119875
21198941198981198611198911198941198981198621199101198941198752119894119898119860
119891119894119898
] (65)
Similarly we have
119875119894119898119861119894= [
1198751119894119898119860
120591119894+ 119875
2119894119898119861119891119894119898119862119910120591119894
1198752119894119898119860
120591119894+ 119875
2119894119898119861119891119894119898119862119910120591119894
]
119875119894119898119863
119894= [
1198751119894119898119863
119894+ 119875
2119894119898119861119891119894119898119863
119910119894
1198752119894119898119863
119894+ 119875
2119894119898119861119891119894119898119863
119910119894
]
(66)
Define 119860119891119894119898
= 1198752119894119898119860
119891119894119898 119861
119891119894119898= 119875
2119894119898119861119891119894119898
119862119891119894119898
= 119862119891119894119898
119872
119860119894119898= 119872
2119894119898119860
119891119894119898119872
119861119894119898= 119872
2119894119898119861119891119894119898
119873119860119894119898
= 1198732119894119898119860
119891119894119898
119873119861119894119898
= 1198732119894119898119861119891119894119898
119881119860119894119898
= 1198812119894119898119860
119891119894119898 and 119881
119861119894119898= 119881
2119894119898119861119891119894119898
Therefore if (59) and (60) hold system (9) is finite-timebounded with a prescribed 119867
infinperformance index 120574 The
proof is completed
Remark 16 In many actual applications the minimum valueof 1205742min is of interest In Theorem 12 with a fixed 120582 120574min canbe obtained through the following optimization procedure
min 1205742
st (48)ndash(50) (67)
In Theorem 15 as for finite-time stability and bounded-ness once the state bound 119888
2is not ascertained theminimum
value 1198882min is of interest With a fixed 120582 define 120582
1= 1 then
the following optimization problem can be formulated to getminimum value 119888
2min
min 1205891205742
+ (1 minus 120589) 1198882
st (59)-(60) and (50)
(68)
where 120589 is weighted factor and 120589 isin [0 1]
5 Illustrative Example
Example 17 Consider that the Markovian jump system andthe delay mode switching are governed by a Markov processwith the following transition rates
Λ = [
minus07 07
09 minus09] Π = [
minus1 1
12 minus12] (69)
as well as with the following parameters
1198601= [
minus35 086
minus064 minus325] 119860
2= [
minus25 034
14 minus002]
1198601205911= [
minus08 minus13
minus07 minus22] 119860
1205912= [
minus28 05
minus08 minus14]
1198631= 119863
2= [
01
02] 119862
1199101= [04 07]
1198621199102= [05 08] 119862
1199101205911= [01 02]
1198621199101205912
= [01 01]
1198631199101= 119863
1199102= 01 119862
1199111= [01 005]
1198621199112= [02 009] 119862
1199111205911= [03 06]
1198621199111205912
= [04 08] 1198631199111= 119863
1199112= 005
(70)
Journal of Applied Mathematics 11
Then we choose 119877 = 119868 119879 = 2 1198881= 1 and 119889 = 001
throughTheorem 15 it yields that themode-dependent filtersare as follows
11986011989111
= [
minus61254 15565
minus02347 minus03763]
11986011989112
= [
minus62682 14338
minus03552 minus05327]
11986011989121
= [
minus89214 24223
minus15476 minus06234]
11986011989122
= [
minus84638 28237
minus13545 minus04322]
11986111989111
= [
51465 minus28516
minus91216 95171]
11986111989112
= [
54203 minus23121
minus93156 96332]
11986111989121
= [
46193 minus11984
minus162397 163006]
11986111989122
= [
46512 minus10311
minus161132 162312]
11986211989111
= [00294 minus00397]
11986211989112
= [00271 minus003368]
11986211989121
= [01163 minus01432]
11986211989122
= [01342 minus03672]
(71)
This paper deals with the finite-time filter design problemfor a class of Markovian jump systems particularly two dif-ferentMarkov processes are considered formodeling the ran-domness of system matrix and the state delay Then throughthe numerical example we can see that results in this paperare feasible which further verified the correctness of our the-ory Therefore the paper shorten this gap
6 Conclusions
In this paper we have examined the problems of finite-time119867
infinfiltering for a class of Markovian jump systems with dif-
ferent system modes Based on a novel approach a sufficientcondition is derived such that the closed-loop Markovianjump system is finite-time bounded and satisfies a prescribedlevel of119867
infindisturbance attenuation in a finite time interval
Finally a numerical example is also given to illustrate theeffectiveness of the proposed design approach It should benoted that one of future research topics would be to investi-gate the problems of fault detection and fault tolerant controlfor time-varying Markovian jump systems with incompleteinformation over a finite-time horizon
Acknowledgments
The authors would like to thank the associate editor and theanonymous reviewers for their detailed comments and sug-gestions This work was supported by the Fund of SichuanProvincial Key Laboratory of Signal and Information Pro-cessing Xihua University (SZJJ2009-002 and SGXZD0101-10-1) and National Basic Research Program of China(2010CB732501)
References
[1] N N Krasovskiı and E A Lidskiı ldquoAnalytical design ofcontrollers in systems with random attributesmdashI Statementof the problem method of solvingrdquo Automation and RemoteControl vol 22 pp 1021ndash1025 1961
[2] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[3] L Wu X Su and P Shi ldquoSliding mode control with bounded1198712gain performance of Markovian jump singular time-delay
systemsrdquo Automatica vol 48 no 8 pp 1929ndash1933 2012[4] H Gao Z Fei J Lam and B Du ldquoFurther results on expo-
nential estimates of Markovian jump systems with mode-de-pendent time-varying delaysrdquo IEEE Transactions on AutomaticControl vol 56 no 1 pp 223ndash229 2011
[5] Z Wu H Su and J Chu ldquo119867infinfiltering for singular Markovian
jump systems with time delayrdquo International Journal of Robustand Nonlinear Control vol 20 no 8 pp 939ndash957 2010
[6] P Tino M Cernansky and L Benuskova ldquoMarkovian architec-tural bias of recurrent neural networksrdquo IEEE Transactions onNeural Networks vol 15 no 1 pp 6ndash15 2004
[7] H Dong Z Wang D W C Ho and H Gao ldquoRobust 119867infin
filtering for Markovian jump systems with randomly occurringnonlinearities and sensor saturation the finite-horizon caserdquoIEEE Transactions on Signal Processing vol 59 no 7 pp 3048ndash3057 2011
[8] Q Ma S Xu and Y Zou ldquoStability and synchronization forMarkovian jump neural networks with partly unknown tran-sition probabilitiesrdquo Neurocomputing vol 74 pp 3404ndash33412011
[9] H Zhao S Xu and Y Zou ldquoRobust119867infinfiltering for uncertain
Markovian jump systems with mode-dependent distributeddelaysrdquo International Journal of Adaptive Control and SignalProcessing vol 24 no 1 pp 83ndash94 2010
[10] Y Zhang S Xu and B Zhang ldquoDiscrete-time fuzzy Markovianjump systems with time-varying delaysrdquo IEEE Transactions onFuzzy Systems vol 17 no 2 pp 411ndash420 2009
[11] Z Wu H Su and J Chu ldquoState estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 no 10ndash12 pp 2247ndash2254 2010
[12] Q Zhu and J Cao ldquoRobust exponential stability of markovianjump impulsive stochastic Cohen-Grossberg neural networkswithmixed time delaysrdquo IEEE Transactions onNeural Networksvol 21 no 8 pp 1314ndash1325 2010
[13] H Huang G Feng and X Chen ldquoStability and stabilizationof Markovian jump systems with time delay via new Lyapunovfunctionalsrdquo IEEE Transactions on Circuits and Systems I vol59 no 10 pp 2413ndash2421 2012
12 Journal of Applied Mathematics
[14] Q Zhu and J Cao ldquoStability of Markovian jump neural net-works with impulse control and time varying delaysrdquoNonlinearAnalysis Real World Applications vol 13 no 5 pp 2259ndash22702012
[15] M Sun J Lam S Xu andY Zou ldquoRobust exponential stabiliza-tion for Markovian jump systems with mode-dependent inputdelayrdquo Automatica vol 43 no 10 pp 1799ndash1807 2007
[16] M Mohammadian A Abolmasoumi and H Momeni ldquo119867infin
mode-independent filter design for Markovian jump geneticregulatory networks with time-varying delaysrdquo Neurocomput-ing vol 87 pp 10ndash18 2012
[17] S Boyd L El Ghaoui E Feron and V Balakrishnan LinearMatrix Inequalities in SystemandControlTheory vol 15 of SIAMStudies in Applied Mathematics SIAM Philadelphia Pa USA1994
[18] S Mou H GaoW Qiang and K Chen ldquoNew delay-dependentexponential stability for neural networks with time delayrdquo IEEETransactions on Systems Man and Cybernetics Part B vol 38no 2 pp 571ndash576 2008
[19] O M Kwon J Park S Lee and E Cha ldquoAnalysis on delay-dependent stability for neural networks with time-varyingdelaysrdquo Neurocomputing vol 103 pp 114ndash120 2013
[20] L Hu H Gao andW X Zheng ldquoNovel stability of cellular neu-ral networkswith interval time-varying delayrdquoNeuralNetworksvol 21 no 10 pp 1458ndash1463 2008
[21] H Zhang Z Liu G B Huang and ZWang ldquoNovel weighting-delay-based stability criteria for recurrent neural networks withtime-varying delayrdquo IEEE Transactions on Neural Networks vol21 no 1 pp 91ndash106 2010
[22] Z Zuo C Yang and YWang ldquoA newmethod for stability anal-ysis of recurrent neural networks with interval time-varyingdelayrdquo IEEE Transactions on Neural Networks vol 21 no 2 pp339ndash344 2010
[23] P Park JWKo andC Jeong ldquoReciprocally convex approach tostability of systems with time-varying delaysrdquo Automatica vol47 no 1 pp 235ndash238 2011
[24] D Zhang L Yu and W Zhang ldquoDelay-dependent fault detec-tion for switched linear systems with time-varying delays-theaverage dwell time approachrdquo Signal Processing vol 91 pp 832ndash840 2011
[25] Z Fei H Gao and P Shi ldquoNew results on stabilization ofMark-ovian jump systems with time delayrdquoAutomatica vol 45 no 10pp 2300ndash2306 2009
[26] H R Karimi ldquoRobust delay-dependent 119867infin
control of uncer-tain time-delay systems with mixed neutral discrete anddistributed time-delays and Markovian switching parametersrdquoIEEE Transactions on Circuits and Systems I vol 58 no 8 pp1910ndash1923 2011
[27] M Luo G Liu and S Zhong ldquoRobust fault detection ofMarkovian jump systems with different systemmodesrdquoAppliedMathematical Modelling vol 37 no 7 pp 5001ndash5012 2013
[28] B Zhang andY Li ldquoExponential1198712minus119871
infinfiltering for distributed
delay systems with Markovian jumping parametersrdquo SignalProcessing vol 93 no 1 pp 206ndash216 2013
[29] A Abolmasoumi and H Momeni ldquoRobust observer-based119867infin
control of a Markovian jump system with different delay andsystem modesrdquo International Journal of Control Automationand Systems vol 9 no 4 pp 768ndash776 2011
[30] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[31] S Ma and E K Boukas ldquoRobust 119867infin
filtering for uncertaindiscrete Markov jump singular systems with mode-dependenttime delayrdquo IET Control Theory amp Applications vol 3 no 3 pp351ndash361 2009
[32] Y ChenW Bi andW Li ldquoStability analysis for neural networkswith time-varying delay a more general delay decompositionapproachrdquo Neurocomputing vol 73 no 4ndash6 pp 853ndash857 2010
[33] Z Shu J Lam and S Xu ldquoRobust stabilization of Markoviandelay systems with delay-dependent exponential estimatesrdquoAutomatica vol 42 no 11 pp 2001ndash2008 2006
[34] X Huang W Lin and B Yang ldquoGlobal finite-time stabilizationof a class of uncertain nonlinear systemsrdquo Automatica vol 41no 5 pp 881ndash888 2005
[35] C Qian and J Li ldquoGlobal finite-time stabilization by outputfeedback for planar systems without observable linearizationrdquoIEEE Transactions on Automatic Control vol 50 no 6 pp 885ndash890 2005
[36] F Amato R Ambrosino C Cosentino and G De TommasildquoInput-output finite time stabilization of linear systemsrdquo Auto-matica vol 46 no 9 pp 1558ndash1562 2010
[37] Y Hong ldquoFinite-time stabilization and stabilizability of a classof controllable systemsrdquo Systems amp Control Letters vol 46 no4 pp 231ndash236 2002
[38] F Amato and M Ariola ldquoFinite-time control of discrete-timelinear systemsrdquo IEEETransactions onAutomatic Control vol 50no 5 pp 724ndash729 2005
[39] Z Zuo Y Liu Y Wang and H Li ldquoFinite-time stochasticstability and stabilisation of linear discrete-time Markovianjump systems with partly unknown transition probabilitiesrdquoIET Control Theory amp Applications vol 6 no 10 pp 1522ndash15262012
[40] X Luan F Liu and P Shi ldquoFinite-time filtering for non-linearstochastic systems with partially known transition jump ratesrdquoIET Control Theory amp Applications vol 4 no 5 pp 735ndash7452010
[41] F Amato M Ariola and C Cosentino ldquoFinite-time control ofdiscrete-time linear systems analysis and design conditionsrdquoAutomatica vol 46 no 5 pp 919ndash924 2010
[42] X Lin H Du and S Li ldquoFinite-time boundedness and 1198712-
gain analysis for switched delay systems with norm-boundeddisturbancerdquo Applied Mathematics and Computation vol 217no 12 pp 5982ndash5993 2011
[43] J Lin S Fei and Z Gao ldquoStabilization of discrete-time switchedsingular time-delay systems under asynchronous switchingrdquoJournal of the Franklin Institute vol 349 no 5 pp 1808ndash18272012
[44] H Song L Yu D Zhang and W-A Zhang ldquoFinite-time 119867infin
control for a class of discrete-time switched time-delay systemswith quantized feedbackrdquoCommunications in Nonlinear Scienceand Numerical Simulation vol 17 no 12 pp 4802ndash4814 2012
[45] Z Zuo H Li Y Liu and Y Wang ldquoOn finite-time stochasticstability and stabilization ofMarkovian jump systems subject topartial information on transition probabilitiesrdquoCircuits Systemsand Signal Processing 2012
[46] H Liu Y Shen and X Zhao ldquoDelay-dependent observer-based119867
infinfinite-time control for switched systems with time-varying
delayrdquoNonlinear Analysis Hybrid Systems vol 6 no 3 pp 885ndash898 2012
[47] Y Zhang C Liu and X Mu ldquoRobust finite-time119867infincontrol of
singular stochastic systems via static output feedbackrdquo AppliedMathematics and Computation vol 218 no 9 pp 5629ndash56402012
Journal of Applied Mathematics 13
[48] J Cheng H Zhu S Zhong Y Zhang and Y Zeng ldquoFinite-timeStabilization of 119867
infinfiltering for switched stochastic systemsrdquo
Circuits Systems and Signal Processing 2012[49] J Cheng H Zhu S Zhong and Y Zhang ldquoFinite-time bound-
ness of 119867infin
filtering for switching discrete-time systemsrdquo In-ternational Journal of Control Automation and Systems vol 10pp 1129ndash1135 2012
[50] Z Xiang C Qiao and M S Mahmoud ldquoFinite-time analysisand119867
infincontrol for switched stochastic systemsrdquo Journal of the
Franklin Institute vol 349 no 3 pp 915ndash927 2012
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Journal of Applied Mathematics 5
Then for each 119903119905= 119894 119904
119905= 119898 we have
pound1198811(120578
119905 119894 119898)
= limΔrarr0
+
1
Δ
E 120578⊺
(119905 + Δ) 119890120575(119905+Δ)
119875119903119905+Δ
119904119905+Δ
120578 (119905 + Δ)
minus120578⊺
(119905) 119890120575119905
119875119894119898120578 (119905)
= limΔrarr0
+
1
Δ
E
120578⊺
(119905 + Δ)[
[
(1 + 120582119894119894Δ + 119900 (Δ))
times (1 + 120587119898119898Δ + 119900 (Δ)) 119890
120575(119905+Δ)
119875119894119898
+ (1 + 120582119894119894Δ + 119900 (Δ))
times ( sum
119899isinM
(120587119898119899Δ + 119900 (Δ))) 119890
120575(119905+Δ)
119875119894119899
+ (1 + 120587119898119898Δ + 119900 (Δ))
times (sum
119895isinN
(120582119894119895Δ + 119900 (Δ))) 119890
120575(119905+Δ)
119875119895119898
+ ( sum
119899isinM
(120587119898119899Δ + 119900 (Δ)))
times(sum
119895isinN
(120582119894119895Δ + 119900 (Δ))) 119890
120575(119905+Δ)
119875119895119899
]
]
times 120578 (119905 + Δ)
minus 120578⊺
(119905) 119890120575119905
119875119894119898120578 (119905)
= 120575119890120575119905
120578⊺
(119905) 119875119894119898120578 (119905) + 2119890
120575119905
120578⊺
(119905) 119875119894119898120578 (119905)
+ 119890120575119905
120578⊺
(119905)( sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898)120578 (119905)
= 119890120575119905
120578⊺
(119905)( sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898+ 120575119875
119894119898
+ 119875119894119898119860
119894+ 119860
⊺
119894119875119894119898)120578 (119905)
+ 2119890120575119905
120578⊺
(119905) 119875119894119898119861119894120578 (119905 minus 120591
119898(119905)) + 2119890
120575119905
120578⊺
(119905) 119875119894119898119863
119894120596 (119905)
(27)pound119881
2(120578
119905 119894 119898)
= limΔrarr0
+
1
Δ
Eint119905+Δ
119905+Δminusℎ
120578⊺
(119904) 119890120575(119904+ℎ)
1198761119903119905+Δ
120578 (119904) 119889119904
minusint
119905
119905minusℎ
120578⊺
(119904) 119890120575(119904+ℎ)
1198761119894120578 (119904) 119889119904
+ limΔrarr0
+
1
Δ
Eint119905+Δ
119905+Δminus120591119904119905+Δ
(119905)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119903119905+Δ
120578 (119904) 119889119904
minusint
119905
119905minus120591119898(119905)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119894120578 (119904) 119889119904
+ limΔrarr0
+
1
Δ
Eint0
minusℎ
int
119905+Δ
119905+Δ+120579
119890120575(119904+ℎ)
120578⊺
(119904) 119876120578 (119904) 119889119904 119889120579
minusint
0
minusℎ
int
119905
119905+120579
119890120575(119904+ℎ)
120578⊺
(119904) 119876120578 (119904) 119889119904 119889120579
= limΔrarr0
+
1
Δ
Eint119905+Δ
119905+Δminusℎ
120578⊺
(119904) 119890120575(119904+ℎ)
1198761119894120578 (119904) 119889119904
+ int
119905+Δ
119905+Δminusℎ
120578⊺
(119904)
times sum
119895isinN
(120582119894119895Δ + 119900 (Δ))
times 119890120575(119904+ℎ)
1198761119894120578 (119904) 119889119904
minusint
119905
119905minusℎ
120578⊺
(119904) 1198761119894120578 (119904) 119889119904
+ limΔrarr0
+
1
Δ
E[int119905+Δ
119905+Δminus120591119895(119905+Δ)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119894120578 (119904) 119889119904
+ sum
119899isinM
(120587119898119899Δ + 119900 (Δ))
times int
119905
119905+Δminus120591119899(119905+Δ)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119894120578 (119904) 119889119904
+ int
119905+Δ
minus120591119895(119905 + Δ)
119905+Δ
120578⊺
(119904)
times sum
119895isinN
(120582119894119895Δ + 119900 (Δ))
times 119890120575(119904+ℎ)
1198762119895120578 (119904) 119889119904
minusint
119905
119905minus120591119898(119905)
120578⊺
(119904) 119890120575(119904+ℎ)
1198762119894120578 (119904) 119889119904]
+ 119890120575119905
120578⊺
(119905) ℎ119876120578 (119905) minus int
119905minusℎ
119890120575(119904+ℎ)
120578⊺
(119904) 119876120578 (119904) 119889119904
le 119890120575119905
120578⊺
(119905) (119890120575ℎ
1198761119894+ 119890
120575ℎ
1198762119894+ ℎ119876) 120578 (119905)
minus 119890120575119905
120578⊺
(119905 minus ℎ)1198761119894120578 (119905 minus ℎ)
minus (1 minus 120591119898(119905)) 119890
120575119905
120578⊺
(119905 minus 120591119898(119905)) 119876
2119894120578 (119905 minus 120591
119898(119905))
+ int
119905
119905minusℎ
119890120590(119904+ℎ)
120578⊺
(119904)( sum
119895isinN119895 = 119894
120582119894119895(119876
1119895+ 119876
2119895)
6 Journal of Applied Mathematics
+ sum
119899isinM119898 = 119899
120587119898119899119876
2119894minus 119876)
times 120578 (119904) 119889119904
(28)
Since
0 le 120591119898(119905) le ℎ
119898 (29)
we define
119903 (120591119898) =
minus (1 minus 120583119898) 119890
120575ℎ119898 if 120583
119898gt 1
minus (1 minus 120583119898) if 120583
119898le 1
(30)
Then
pound1198812(120578
119905 119894 119898) le 119890
120575119905
120578⊺
(119905) (119890120575ℎ
1198761119894+ 119890
120575ℎ
1198762119894+ ℎ119876) 120578 (119905)
minus 119890120575119905
120578⊺
(119905 minus ℎ)1198761119894120578 (119905 minus ℎ)
+ 119890120575119905
120578⊺
(119905 minus 120591119898(119905)) 119903 (120591
119898) 119876
2119894120578 (119905 minus 120591
119898(119905))
+ int
119905
119905minusℎ
119890120590(119904+ℎ)
120578⊺
(119904)
times ( sum
119895isinN119895 = 119894
120582119894119895(119876
1119895+ 119876
2119895)
+ sum
119899isinM119898 = 119899
120587119898119899119876
2119894minus 119876)
times 120578 (119904) 119889119904
(31)
Similar to the previous process we can obtain
pound1198813(120578
119905 119894 119898) le int
0
minusℎ
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904)
times ( sum
119899isinM119898 = 119899
120587119898119899119883
119894119899
+ sum
119895isinN119894 = 119895
120582119894119895119883
119895119898minus 119883)
times 120578 (119904) 119889119904 119889120579
+ 119890120575119905
120578⊺
(119905) 119883119894119898120578 (119905) int
0
minusℎ
119890minus120575120592
119889120592
minus 119890120575119905
int
119905
119905minusℎ
120578⊺
(119904)119883119894119898120578 (119904) 119889119904
+ 119890120575119905
120578⊺
(119905) 119883120578 (119905) int
0
minusℎ
int
0
120592
119890minus120575120592
119889120579119889120592
(32)
By using Lemma 5 it yields that
minusint
119905
119905minusℎ
120578⊺
(119904)119883119894119898120578 (119904) 119889119904 = minusint
119905
119905minus120591119898(119905)
120578⊺
(119904)119883119894119898120578 (119904) 119889119904
minus int
119905minus120591119898(119905)
119905minusℎ
120578⊺
(119904)119883119894119898120578 (119904) 119889119904
le minus120591119898(119905) 119880
⊺
1119883
1198941198981198801
minus (ℎ minus 120591119898(119905)) 119880
⊺
2119883
1198941198981198802
(33)
where
1198801=
1
120591119898(119905)
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904
1198802=
1
ℎ minus 120591119898(119905)
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904
lim120591119898(119905)rarr0
1
120591119898(119905)
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904 = 120578 (119905)
lim120591119898(119905)rarrℎ
1
ℎ minus 120591119898(119905)
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904 = 120578 (119905 minus ℎ)
(34)
From the Newton-Leibniz formula the following equa-tion is true for any matrices119872
119894119898 119873
119894119898 and 119881
119894119898with appro-
priate dimensions
(2120591119898(119905) 119880
⊺
1119872
119894119898+ 2 (ℎ minus 120591
119898(119905)) 119880
⊺
2119873
119894119898+ 2 120578
⊺
(119905) 119881119894119898)
times [minus 120578 (119905) + 119860119894120578 (119905) + 119861
120591119894120578 (119905 minus 120591
119898(119905)) + 119863
119894120596 (119905)] = 0
(35)
pound1198814(120578
119905 119894 119898)
le int
0
minusℎ
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904)
times ( sum
119899isinM119898 = 119899
120587119898119899119884119894119899
+ sum
119895isinN119894 = 119895
120582119894119895119884119895119898minus 119884)
times 120578 (119904) 119889119904 119889120579
+ 119890120575119905
120578⊺
(119905) 119884119894119898120578 (119905) int
0
minusℎ
119890minus120575120592
119889120592
minus 119890120575119905
int
119905
119905minusℎ
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
+ 119890120575119905
120578⊺
(119905) 119884 120578 (119905) int
0
minusℎ
int
0
120592
119890minus120575120592
119889120579 119889120592
(36)
Journal of Applied Mathematics 7
From Lemma 5 it yields that
minus int
119905
119905minusℎ
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
= minusint
119905
119905minus120591119898(119905)
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
minus int
119905minus120591119898(119905)
119905minusℎ
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
le minus
ℎ
120591119898(119905)
[int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904]
⊺
119884119894119898
ℎ
[int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904]
minus
ℎ
ℎ minus 120591119898(119905)
[int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904]
⊺
119884119894119898
ℎ
[int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904]
le minus
[
[
[
[
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904
]
]
]
]
⊺
[
[
[
119884119894119898
ℎ
119878119894119898
lowast
119884119894119898
ℎ
]
]
]
[
[
[
[
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904
]
]
]
]
(37)
From (25)ndash(37) we can eventually obtain
pound119881 (120578119905 119903
119905 119904
119905) minus 120575120596
⊺
(119905)119867120596 (119905) le 119890120575119905
120585⊺
(119905) Ξ119894119898120585 (119905) (38)
where
120585⊺
(119905) = [120578⊺
(119905) 120578⊺
(119905 minus 120591119898(119905)) 120578
⊺
(119905 minus ℎ) 120578⊺
(119905) 119880⊺
1 119880
⊺
2 120596
⊺
(119905)]
Ξ119894119898=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860⊺
119894119881
⊺
119894119898120591119898(119905) 119860
⊺
119894119872
⊺
119894119898(ℎ minus 120591
119898(119905)) 119860
⊺
119894119873
⊺
119894119898119875119894119898119863
119894
lowast Ξ22119894119898
119878119894119898minus
119884119894119898
ℎ
119860⊺
120591119894119881
⊺
119894119898120591119898(119905) 119860
⊺
120591119894119872
⊺
119894119898(ℎ minus 120591
119898(119905)) 119860
⊺
120591119894119873
⊺
1198941198980
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minus120591119898(119905)119872
⊺
119894119898minus (ℎ minus 120591
119898(119905))119873
⊺
119894119898119881119894119898119863
119894
lowast lowast lowast lowast minus120591119898(119905) 119883
1198941198980 120591
119898(119905)119872
119894119898119863
119894
lowast lowast lowast lowast lowast minus (ℎ minus 120591119898(119905))119883
119894119898(ℎ minus 120591
119898(119905))119873
119894119898119863
119894
lowast lowast lowast lowast lowast lowast minus120575119867
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
(39)
The LMIs (20) and (21) lead to 120591119898(119905) rarr ℎ and 120591
119898(119905) rarr
0 respectively It is easy to see that Ξ1119894119898
results from Ξ119894119898|
120591119898(119905) = ℎ and Ξ
119894119898|120591119898(119905) = 0 Thus we obtain
E pound119881 (120578119905 119903
119905 119904
119905) le 120578E [119881 (120578
119905 119903
119905 119904
119905)] + 120575120596
⊺
(119905)119867120596 (119905)
(40)
Multiplying the aforementioned inequality by 119890minus120578119905 we canget
E pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] le 119890
minus120578119905
120575120596⊺
(119905)119867120596 (119905) (41)
By integrating the aforementioned inequality between 0and 119905 it follows that
119890minus120578119905
E [119881 (120578119905 119903
119905 119904
119905)] minus E [119881 (120578
0 119903
0 119904
0)]
le 120575int
119905
0
119890minus120578119904
120596⊺
(119904)119867120596 (119904) 119889119904
(42)
Denote that 119894119898
= 119877minus(12)
119875119894119898119877minus(12) 119876
119904119894=
119877minus(12)
119876119904119894119877minus(12)
(119904 = 1 2) 119876 = 119877minus(12)
119876119877minus(12)
119883119894119898
= 119877minus(12)
119883119894119898119877minus(12) 119883 = 119877
minus(12)
119883119877minus(12)
119894119898
= 119877minus(12)
119884119894119898119877minus(12) and = 119877
minus(12)
119884119877minus(12) it yields
that
E [119881 (1205780 119903
0 119904
0)]
le max119894isinN119898isinM
120582max (119894119898) 120578⊺
(0) 119877120578 (0)
+ (max119894isinN
120582max (1198761119894) +max
119894isinN120582max (1198762119894
)) 119890120575ℎ
times int
0
minusℎ
119890120575119904
120578⊺
(119904) 119877120578 (119904) 119889119904
+ 119890120575ℎ
120582max (119876)int0
minusℎ
int
0
120579
119890120575119904
120578⊺
(119904) 119877120578 (119904) 119889119904
+ 119890120575ℎ max
119894isinN119898isinM120582max (119883119894119898
)
times int
0
minusℎ
int
0
120579
119890minus120575120579
120578⊺
(119904) 119877120578 (119904) 119889119904 119889120579
+ 119890120575ℎ
120582max (119883)
times int
0
minusℎ
int
0
120579
int
0
120592
119890minus120575120579
120578⊺
(119904) 119877120578 (119904) 119889119904 119889120579 119889120592
+ 119890120575ℎ max
119894isinN119898isinM120582max (119894119898)
8 Journal of Applied Mathematics
times int
0
minusℎ
int
0
120579
119890minus120575120579
120578⊺
(119904) 119877 120578 (119904) 119889119904 119889120579
+ 119890120575ℎ
120582max ()
times int
0
minusℎ
int
0
120579
int
0
120592
119890minus120575120579
120578⊺
(119904) 119877 120578 (119904) 119889119904 119889120579 119889120592
le max119894isinN119898isinM
120582max (119894119898)
+ ℎ119890120575ℎ
(max119894isinN
120582max (1198761119894+
119894isinNmax120582max (1198762119894
)))
+ ℎ2
119890120575ℎ
( max119894isinN119898isinM
120582max (119883119894119898) + 120582max (119876)
+ max119894isinN119898isinM
120582max (119894119898))
+
1
2
ℎ3
119890120575ℎ
(120582max (119883) + 120582max ())
times supminusℎle119904le0
120578⊺
(119904) 119877120578 (119904) 120578⊺
(119904) 119877 120578 (119904)
= 1198881Λ
(43)
For given 120578 gt 0 and 0 le 119905 le 119879 we have
E [119881 (120578119905 119903
119905 119904
119905)] le E [119890
120578119905
119881 (1205780 119903
0 119904
0)]
+ 119890120578119905
120575int
119905
0
119890minus120578119904
120596⊺
(119904)119867120596 (119904) 119889119904
le 119890120578119879
1198881Λ + 119889120575119890
120578119879
120582max (119867)int119879
0
119890minus120578119904
119889119904
le 119890120578119879
1198881Λ + 119889120575120582
10
1
120578
(1 minus 119890minus120578119879
)
(44)
On the other hand it follows from (25) that
E [119881 (120578119905 119903
119905 119904
119905)] ge E [120578
⊺
(119905) 119890120582119905
119875119894119898120578 (119905)]
ge max119894isinN
120582min (119875119894119898)E [120578⊺
(119905) 119877120578 (119905)]
= 1205821E [120578
⊺
(119905) 119877120578 (119905)]
(45)
It can be derived from (43)ndash(45) that
E [120578⊺
(119905) 119877120578 (119905)] le
119890120578119879
1205821
1198881Λ + 119889120575120582
10
1
120578
(1 minus 119890minus120578119879
) (46)
From (22) and (46) we have
E [120578⊺
(119905) 119877120578 (119905)] lt 1198882 (47)
Then the system is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879)
Remark 8 In this paper 120591119898(119905) and 120591
119898(119905) may have different
upper bounds in various delay intervals satisfying (7) respec-tively However in previous work such as [20 21] 120591
119898(119905) and
120591119898(119905) are enlarged to 120591
119898(119905) le ℎ = maxℎ
119894 119898 isin M and
120591119898(119905) le 120583 = max120583
119898 119898 isin M respectively which may lead
to conservativeness inevitably However the previouse casecan be taken fully into account by employing the Lyapunov-Krasovskii functional (25)
Remark 9 When dealing with termminusint
119905
119905minusℎ
120578⊺
(119904)119884119894119898120578(119904)119889119904 the
convex combination is not employed Lemma 5 is used in thispaper and then the free-weighting matrices-dependent nulladd items are necessary to be introduced in our proof whichlead to the decrease of the number of LMIs and LMIs scalardecision variables
Remark 10 The feature of this paper is the way to dealwith the integral term Many researchers have enlarged thederivative of the Lyapunov functional in order to deal with theintegral term in mathematical operations In this paper wepropose a novel delay-dependent sufficient criterion whichensures that the Markovian jump system with different modesystems is finite-time stable
Remark 11 It should be pointed out that the novelty of theLyapunov functional (25) lies in distinct Lyapunov matrices(119875
119894119898 119883
119894119898 119884
119894119898) which is chosen for different system modes
119894 (119894 = 1 2 119873) and119898 (119898 = 1 2 119872)
Theorem 12 System (9) is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879) if there exist matrices 119875
119894gt 0 119876
119897119894gt 0 (119897 = 1 2)
119876 gt 0 119883119894119898gt 0 119883 gt 0 119884
119894119898gt 0 119884 gt 0 119878
119894119898 scalars 119888
1lt 119888
2
119879 gt 0 120582119904gt 0 (119904 = 1 2 9) 120578 gt 0 120574 gt 0 and Λ gt 0 such
that for all 119894 119895 isinN and119898 119899 isinM and (16)ndash(19) the followinginequalities hold
Σ1119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119872⊺
119894119898119875119894119898119863119894
119862
⊺
119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119872⊺
1198941198980 119864
⊺
119894
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119872⊺
119894119898119881119894119898119863119894
0
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119872119894119898119863119894
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(48)
Σ2119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119873⊺
119894119898119875119894119898119863119894
119862
⊺
119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119873⊺
1198941198980 119864
⊺
119894
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119873⊺
119894119898119881119894119898119863119894
0
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119873119894119898119863119894
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(49)
1198881Λ + 119889120575120574
21
120578
(1 minus 119890minus120578119879
) lt 1205821119890minus120578119879
1198882 (50)
Journal of Applied Mathematics 9
Proof We now consider the 119867infin
performance of system (9)Select the same Lyapunov-Krasovskii functional as Theo-rem 7 it yields that
pound119881 (120578119905 119903
119905 119904
119905) + 119890
⊺
(119905) 119890 (119905) minus 1205742
120596⊺
(119905) 120596 (119905) le 120585⊺
(119905) Σ119897119894119898120585 (119905)
(119897 = 1 2)
(51)
It follows from (49)-(50) that
E pound119881 (120578119905 119903
119905 119904
119905)
le E [120578119881 (120578119905 119903
119905 119904
119905)] + 120574
2
120596⊺
(119905) 120596 (119905) minus E [119890⊺
(119905) 119890 (119905)]
(52)
Multiplying the aforementioned inequality by 119890minus120578119905 onehas
E pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] le 119890
minus120578119905
[1205742
120596⊺
(119905) 120596 (119905) minus 119890⊺
(119905) 119890 (119905)]
(53)
In zero initial condition and E[119881(120578119905 119903
119905 119904
119905)] gt 0 by inte-
grating the aforementioned inequality between 0 and 119879 wecan get
int
119879
0
119890minus120578119905
[1205742
120596⊺
(119905) 120596 (119905) minus 119890⊺
(119905) 119890 (119905)] 119889119905
le Eint119879
0
pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] 119889119905 le 119881 (120578
0 119903
0 119904
0) = 0
(54)
Using Dynkins formula it results that
E [int119879
0
119890minus120578120592
119890⊺
(120592) 119890 (120592) 119889120592] le 1205742
E [int119879
0
119890minus120578120592
120596⊺
(120592) 120596 (120592) 119889120592]
(55)
Then it yields that
E [int119879
0
119890⊺
(120592) 119890 (120592) 119889120592] le 1205742
119890120578119879
E [int119879
0
120596⊺
(120592) 120596 (120592) 119889120592] (56)
Thus it is concluded by Definition 3 that system (9) is fi-nite-time bounded with an 119867
infinperformance 120574 This com-
pletes the proof
Remark 13 From the proof process of Theorems 7 and 12it is easy to see that neither bounding technique for cross-terms nor model transformation is involved In other wordsthe obtained result is expected to be less conservative
Remark 14 The Lyapunov asymptotic stability and finite-time stability of a class of system are independent conceptsA Lyapunov asymptotic stability system may not be finite-time stability Moreover finite-time stability system may alsonot be Lyapunov asymptotic stabilityThere exist some resultson Lyapunov stability while finite-time stability also needsour full investigation which was neglected by most previouswork
4 Finite-Time 119867infin
Filtering
Theorem 15 System (9) is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879) if there exist matrices 119875
119894119898gt 0119872
119894119898 119873
119894119898 119881
119894119898
119876119897119894gt 0 (119897 = 1 2) 119876 gt 0 119883
119894119898gt 0 119883 gt 0 119884
119894119898gt 0 119884 gt 0
119860119891119894119898
119861119891119894119898
119862119891119894119898
119872119860119894119898
119872119861119894119898
119873119860119894119898
119873119861119894119898
119881119860119894119898
119881119861119894119898
119878119894119898
scalars 1198881lt 119888
2 119879 gt 0 120590
119904gt 0 (119904 = 1 2 9) 120575 gt 0 120578 gt 0
and Λ gt 0 such that for all 119894 119895 isin N and 119898 119899 isin M thefollowing inequalities hold
119875119894119898= [
1198751119894119898
1198752119894119898
1198752119894119898
1198752119894119898
] gt 0 119872119894119898= [
1198721119894119898
1198722119894119898
1198722119894119898
1198722119894119898
] (57)
119873119894119898= [
1198731119894119898
1198732119894119898
1198732119894119898
1198732119894119898
] 119881119894119898= [
1198811119894119898
1198812119894119898
1198812119894119898
1198812119894119898
] (58)
Γ1119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Γ11119894119898
Γ12119894119898
minus119878119894119898
Γ14119894119898
Γ15119894119898
Γ16119894119898
Γ17119894119898
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
Γ24119894119898
Γ25119894119898
0 [119862⊺
119911120591119894
0]
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119872⊺
119894119898Γ46119894119898
0
lowast lowast lowast lowast minusℎ119883119894119898
Γ56119894119898
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(59)
Γ2119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Γ11119894119898
Γ12119894119898
minus119878119894119898
Γ14119894119898
Γ1015840
15119894119898Γ16119894119898
Γ17119894119898
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
Γ24119894119898
Γ1015840
251198941198980 [
119862⊺
119911120591119894
0]
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119873⊺
119894119898Γ1015840
461198941198980
lowast lowast lowast lowast minusℎ119883119894119898
Γ1015840
561198941198980
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(60)where
Γ11119894119898
= sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898+ 119890
120575ℎ
1198761119894+ 119890
120575ℎ
1198762119894+ ℎ119876 +
119890120575ℎ
minus 1
120575
119883119894119898+
119890120575ℎ
minus 120575ℎ119890120575ℎ
minus 1
1205752
119883
+
119884119894119898
ℎ
+ [
1205751198751119894119898
+ 1198751119894119898119860
119894+ 119860
⊺
1198941198751119894119898
+ 119861119891119894119898119862119910119894+ 119862
⊺
119910119894119861
⊺
119891119894119898120575119875
2119894119898+ 119860
119891119894119898+ 119860
⊺
119891119894119898
1205751198752119894119898
+ 1198752119894119898119860
119894+ 119860
⊺
1198941198752119894119898
+ 119861119891119894119898119862119910119894+ 119862
⊺
119910119894119861
⊺
119891119894119898120575119875
2119894119898+ 119860
119891119894119898+ 119860
⊺
119891119894119898
]
(61)
10 Journal of Applied Mathematics
Γ12119894119898
= minus
119884119894119898
ℎ
+ 119878119894119898+ [
1198751119894119898119860
120591119894+ 119861
119891119894119898119862119910120591119894
1198752119894119898119860
120591119894+ 119861
119891119894119898119862119910120591119894
]
Γ14119894119898
= [
119860⊺
119894119881
⊺
1119894119898+ 119862
⊺
119910119894119881
⊺
119861119894119898119881
⊺
119860119894119898
119860⊺
119894119881
⊺
2119894119898+ 119862
⊺
119910119894119881
⊺
119861119894119898119881
⊺
119860119894119898
]
Γ15119894119898
= [
ℎ119860⊺
119894119872
⊺
1119894119898+ ℎ119862
⊺
119910119894119872
⊺
119861119894119898ℎ119872
⊺
119860119894119898
ℎ119860⊺
119894119872
⊺
2119894119898+ ℎ119862
⊺
119910119894119872
⊺
119861119894119898ℎ119872
⊺
119860119894119898
]
Γ1015840
15119894119898= [
ℎ119860⊺
119894119873
⊺
1119894119898+ ℎ119862
⊺
119910119894119873
⊺
119861119894119898ℎ119873
⊺
119860119894119898
ℎ119860⊺
119894119873
⊺
2119894119898+ ℎ119862
⊺
119910119894119873
⊺
119861119894119898ℎ119873
⊺
119860119894119898
]
Γ16119894119898
= [
1198751119894119898119863
119894+ 119861
119891119894119898119863
119910119894
1198752119894119898119863
119894+ 119861
119891119894119898119863
119910119894
]
Γ17119894119898
= [
119862⊺
119911119894
minus119862119891119894119898
]
Γ24119894119898
= [
119860⊺
120591119894119881
⊺
1119894119898+ 119862
⊺
119910120591119894119881
⊺
119861119894119898
119860⊺
120591119894119881
⊺
2119894119898+ 119862
⊺
119910120591119894119881
⊺
119861119894119898
]
Γ25119894119898
= [
ℎ119860⊺
120591119894119872
⊺
1119894119898+ ℎ119862
⊺
119910120591119894119872
⊺
119861119894119898
ℎ119860⊺
120591119894119872
⊺
2119894119898+ ℎ119862
⊺
119910120591119894119872
⊺
119861119894119898
]
Γ1015840
25119894119898= [
ℎ119860⊺
120591119894119873
⊺
1119894119898+ ℎ119862
⊺
119910120591119894119873
⊺
119861119894119898
ℎ119860⊺
120591119894119873
⊺
2119894119898+ ℎ119862
⊺
119910120591119894119873
⊺
119861119894119898
]
Γ46119894119898
= [
1198811119894119898119863
119894+ 119881
119861119894119898119863
119910119894
1198812119894119898119863
119894+ 119881
119861119894119898119863
119910119894
]
Γ56119894119898
= [
ℎ1198721119894119898119863
119894+ ℎ119872
119861119894119898119863
119910119894
ℎ1198722119894119898119863
119894+ ℎ119872
119861119894119898119863
119910119894
]
Γ1015840
56119894119898= [
ℎ1198731119894119898119863
119894+ ℎ119873
119861119894119898119863
119910119894
ℎ1198732119894119898119863
119894+ ℎ119873
119861119894119898119863
119910119894
]
(62)
Then a desired filter can be chosen with parameters as
119860119891119894119898
= 119875minus1
2119894119898119860
119891119894119898 119861
119891119894119898= 119875
minus1
2119894119898119861119891119894119898
119862119891119894119898
= 119862119891119894119898
(63)
Proof We denote that
119875119894119898= [
1198751119894119898
1198752119894119898
1198752119894119898
1198752119894119898
] (64)
The term 119875119894119898119860
119894can be rewritten as
119875119894119898119860
119894= [
1198751119894119898119860
119894+ 119875
21198941198981198611198911198941198981198621199101198941198752119894119898119860
119891119894119898
1198752119894119898119860
119894+ 119875
21198941198981198611198911198941198981198621199101198941198752119894119898119860
119891119894119898
] (65)
Similarly we have
119875119894119898119861119894= [
1198751119894119898119860
120591119894+ 119875
2119894119898119861119891119894119898119862119910120591119894
1198752119894119898119860
120591119894+ 119875
2119894119898119861119891119894119898119862119910120591119894
]
119875119894119898119863
119894= [
1198751119894119898119863
119894+ 119875
2119894119898119861119891119894119898119863
119910119894
1198752119894119898119863
119894+ 119875
2119894119898119861119891119894119898119863
119910119894
]
(66)
Define 119860119891119894119898
= 1198752119894119898119860
119891119894119898 119861
119891119894119898= 119875
2119894119898119861119891119894119898
119862119891119894119898
= 119862119891119894119898
119872
119860119894119898= 119872
2119894119898119860
119891119894119898119872
119861119894119898= 119872
2119894119898119861119891119894119898
119873119860119894119898
= 1198732119894119898119860
119891119894119898
119873119861119894119898
= 1198732119894119898119861119891119894119898
119881119860119894119898
= 1198812119894119898119860
119891119894119898 and 119881
119861119894119898= 119881
2119894119898119861119891119894119898
Therefore if (59) and (60) hold system (9) is finite-timebounded with a prescribed 119867
infinperformance index 120574 The
proof is completed
Remark 16 In many actual applications the minimum valueof 1205742min is of interest In Theorem 12 with a fixed 120582 120574min canbe obtained through the following optimization procedure
min 1205742
st (48)ndash(50) (67)
In Theorem 15 as for finite-time stability and bounded-ness once the state bound 119888
2is not ascertained theminimum
value 1198882min is of interest With a fixed 120582 define 120582
1= 1 then
the following optimization problem can be formulated to getminimum value 119888
2min
min 1205891205742
+ (1 minus 120589) 1198882
st (59)-(60) and (50)
(68)
where 120589 is weighted factor and 120589 isin [0 1]
5 Illustrative Example
Example 17 Consider that the Markovian jump system andthe delay mode switching are governed by a Markov processwith the following transition rates
Λ = [
minus07 07
09 minus09] Π = [
minus1 1
12 minus12] (69)
as well as with the following parameters
1198601= [
minus35 086
minus064 minus325] 119860
2= [
minus25 034
14 minus002]
1198601205911= [
minus08 minus13
minus07 minus22] 119860
1205912= [
minus28 05
minus08 minus14]
1198631= 119863
2= [
01
02] 119862
1199101= [04 07]
1198621199102= [05 08] 119862
1199101205911= [01 02]
1198621199101205912
= [01 01]
1198631199101= 119863
1199102= 01 119862
1199111= [01 005]
1198621199112= [02 009] 119862
1199111205911= [03 06]
1198621199111205912
= [04 08] 1198631199111= 119863
1199112= 005
(70)
Journal of Applied Mathematics 11
Then we choose 119877 = 119868 119879 = 2 1198881= 1 and 119889 = 001
throughTheorem 15 it yields that themode-dependent filtersare as follows
11986011989111
= [
minus61254 15565
minus02347 minus03763]
11986011989112
= [
minus62682 14338
minus03552 minus05327]
11986011989121
= [
minus89214 24223
minus15476 minus06234]
11986011989122
= [
minus84638 28237
minus13545 minus04322]
11986111989111
= [
51465 minus28516
minus91216 95171]
11986111989112
= [
54203 minus23121
minus93156 96332]
11986111989121
= [
46193 minus11984
minus162397 163006]
11986111989122
= [
46512 minus10311
minus161132 162312]
11986211989111
= [00294 minus00397]
11986211989112
= [00271 minus003368]
11986211989121
= [01163 minus01432]
11986211989122
= [01342 minus03672]
(71)
This paper deals with the finite-time filter design problemfor a class of Markovian jump systems particularly two dif-ferentMarkov processes are considered formodeling the ran-domness of system matrix and the state delay Then throughthe numerical example we can see that results in this paperare feasible which further verified the correctness of our the-ory Therefore the paper shorten this gap
6 Conclusions
In this paper we have examined the problems of finite-time119867
infinfiltering for a class of Markovian jump systems with dif-
ferent system modes Based on a novel approach a sufficientcondition is derived such that the closed-loop Markovianjump system is finite-time bounded and satisfies a prescribedlevel of119867
infindisturbance attenuation in a finite time interval
Finally a numerical example is also given to illustrate theeffectiveness of the proposed design approach It should benoted that one of future research topics would be to investi-gate the problems of fault detection and fault tolerant controlfor time-varying Markovian jump systems with incompleteinformation over a finite-time horizon
Acknowledgments
The authors would like to thank the associate editor and theanonymous reviewers for their detailed comments and sug-gestions This work was supported by the Fund of SichuanProvincial Key Laboratory of Signal and Information Pro-cessing Xihua University (SZJJ2009-002 and SGXZD0101-10-1) and National Basic Research Program of China(2010CB732501)
References
[1] N N Krasovskiı and E A Lidskiı ldquoAnalytical design ofcontrollers in systems with random attributesmdashI Statementof the problem method of solvingrdquo Automation and RemoteControl vol 22 pp 1021ndash1025 1961
[2] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[3] L Wu X Su and P Shi ldquoSliding mode control with bounded1198712gain performance of Markovian jump singular time-delay
systemsrdquo Automatica vol 48 no 8 pp 1929ndash1933 2012[4] H Gao Z Fei J Lam and B Du ldquoFurther results on expo-
nential estimates of Markovian jump systems with mode-de-pendent time-varying delaysrdquo IEEE Transactions on AutomaticControl vol 56 no 1 pp 223ndash229 2011
[5] Z Wu H Su and J Chu ldquo119867infinfiltering for singular Markovian
jump systems with time delayrdquo International Journal of Robustand Nonlinear Control vol 20 no 8 pp 939ndash957 2010
[6] P Tino M Cernansky and L Benuskova ldquoMarkovian architec-tural bias of recurrent neural networksrdquo IEEE Transactions onNeural Networks vol 15 no 1 pp 6ndash15 2004
[7] H Dong Z Wang D W C Ho and H Gao ldquoRobust 119867infin
filtering for Markovian jump systems with randomly occurringnonlinearities and sensor saturation the finite-horizon caserdquoIEEE Transactions on Signal Processing vol 59 no 7 pp 3048ndash3057 2011
[8] Q Ma S Xu and Y Zou ldquoStability and synchronization forMarkovian jump neural networks with partly unknown tran-sition probabilitiesrdquo Neurocomputing vol 74 pp 3404ndash33412011
[9] H Zhao S Xu and Y Zou ldquoRobust119867infinfiltering for uncertain
Markovian jump systems with mode-dependent distributeddelaysrdquo International Journal of Adaptive Control and SignalProcessing vol 24 no 1 pp 83ndash94 2010
[10] Y Zhang S Xu and B Zhang ldquoDiscrete-time fuzzy Markovianjump systems with time-varying delaysrdquo IEEE Transactions onFuzzy Systems vol 17 no 2 pp 411ndash420 2009
[11] Z Wu H Su and J Chu ldquoState estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 no 10ndash12 pp 2247ndash2254 2010
[12] Q Zhu and J Cao ldquoRobust exponential stability of markovianjump impulsive stochastic Cohen-Grossberg neural networkswithmixed time delaysrdquo IEEE Transactions onNeural Networksvol 21 no 8 pp 1314ndash1325 2010
[13] H Huang G Feng and X Chen ldquoStability and stabilizationof Markovian jump systems with time delay via new Lyapunovfunctionalsrdquo IEEE Transactions on Circuits and Systems I vol59 no 10 pp 2413ndash2421 2012
12 Journal of Applied Mathematics
[14] Q Zhu and J Cao ldquoStability of Markovian jump neural net-works with impulse control and time varying delaysrdquoNonlinearAnalysis Real World Applications vol 13 no 5 pp 2259ndash22702012
[15] M Sun J Lam S Xu andY Zou ldquoRobust exponential stabiliza-tion for Markovian jump systems with mode-dependent inputdelayrdquo Automatica vol 43 no 10 pp 1799ndash1807 2007
[16] M Mohammadian A Abolmasoumi and H Momeni ldquo119867infin
mode-independent filter design for Markovian jump geneticregulatory networks with time-varying delaysrdquo Neurocomput-ing vol 87 pp 10ndash18 2012
[17] S Boyd L El Ghaoui E Feron and V Balakrishnan LinearMatrix Inequalities in SystemandControlTheory vol 15 of SIAMStudies in Applied Mathematics SIAM Philadelphia Pa USA1994
[18] S Mou H GaoW Qiang and K Chen ldquoNew delay-dependentexponential stability for neural networks with time delayrdquo IEEETransactions on Systems Man and Cybernetics Part B vol 38no 2 pp 571ndash576 2008
[19] O M Kwon J Park S Lee and E Cha ldquoAnalysis on delay-dependent stability for neural networks with time-varyingdelaysrdquo Neurocomputing vol 103 pp 114ndash120 2013
[20] L Hu H Gao andW X Zheng ldquoNovel stability of cellular neu-ral networkswith interval time-varying delayrdquoNeuralNetworksvol 21 no 10 pp 1458ndash1463 2008
[21] H Zhang Z Liu G B Huang and ZWang ldquoNovel weighting-delay-based stability criteria for recurrent neural networks withtime-varying delayrdquo IEEE Transactions on Neural Networks vol21 no 1 pp 91ndash106 2010
[22] Z Zuo C Yang and YWang ldquoA newmethod for stability anal-ysis of recurrent neural networks with interval time-varyingdelayrdquo IEEE Transactions on Neural Networks vol 21 no 2 pp339ndash344 2010
[23] P Park JWKo andC Jeong ldquoReciprocally convex approach tostability of systems with time-varying delaysrdquo Automatica vol47 no 1 pp 235ndash238 2011
[24] D Zhang L Yu and W Zhang ldquoDelay-dependent fault detec-tion for switched linear systems with time-varying delays-theaverage dwell time approachrdquo Signal Processing vol 91 pp 832ndash840 2011
[25] Z Fei H Gao and P Shi ldquoNew results on stabilization ofMark-ovian jump systems with time delayrdquoAutomatica vol 45 no 10pp 2300ndash2306 2009
[26] H R Karimi ldquoRobust delay-dependent 119867infin
control of uncer-tain time-delay systems with mixed neutral discrete anddistributed time-delays and Markovian switching parametersrdquoIEEE Transactions on Circuits and Systems I vol 58 no 8 pp1910ndash1923 2011
[27] M Luo G Liu and S Zhong ldquoRobust fault detection ofMarkovian jump systems with different systemmodesrdquoAppliedMathematical Modelling vol 37 no 7 pp 5001ndash5012 2013
[28] B Zhang andY Li ldquoExponential1198712minus119871
infinfiltering for distributed
delay systems with Markovian jumping parametersrdquo SignalProcessing vol 93 no 1 pp 206ndash216 2013
[29] A Abolmasoumi and H Momeni ldquoRobust observer-based119867infin
control of a Markovian jump system with different delay andsystem modesrdquo International Journal of Control Automationand Systems vol 9 no 4 pp 768ndash776 2011
[30] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[31] S Ma and E K Boukas ldquoRobust 119867infin
filtering for uncertaindiscrete Markov jump singular systems with mode-dependenttime delayrdquo IET Control Theory amp Applications vol 3 no 3 pp351ndash361 2009
[32] Y ChenW Bi andW Li ldquoStability analysis for neural networkswith time-varying delay a more general delay decompositionapproachrdquo Neurocomputing vol 73 no 4ndash6 pp 853ndash857 2010
[33] Z Shu J Lam and S Xu ldquoRobust stabilization of Markoviandelay systems with delay-dependent exponential estimatesrdquoAutomatica vol 42 no 11 pp 2001ndash2008 2006
[34] X Huang W Lin and B Yang ldquoGlobal finite-time stabilizationof a class of uncertain nonlinear systemsrdquo Automatica vol 41no 5 pp 881ndash888 2005
[35] C Qian and J Li ldquoGlobal finite-time stabilization by outputfeedback for planar systems without observable linearizationrdquoIEEE Transactions on Automatic Control vol 50 no 6 pp 885ndash890 2005
[36] F Amato R Ambrosino C Cosentino and G De TommasildquoInput-output finite time stabilization of linear systemsrdquo Auto-matica vol 46 no 9 pp 1558ndash1562 2010
[37] Y Hong ldquoFinite-time stabilization and stabilizability of a classof controllable systemsrdquo Systems amp Control Letters vol 46 no4 pp 231ndash236 2002
[38] F Amato and M Ariola ldquoFinite-time control of discrete-timelinear systemsrdquo IEEETransactions onAutomatic Control vol 50no 5 pp 724ndash729 2005
[39] Z Zuo Y Liu Y Wang and H Li ldquoFinite-time stochasticstability and stabilisation of linear discrete-time Markovianjump systems with partly unknown transition probabilitiesrdquoIET Control Theory amp Applications vol 6 no 10 pp 1522ndash15262012
[40] X Luan F Liu and P Shi ldquoFinite-time filtering for non-linearstochastic systems with partially known transition jump ratesrdquoIET Control Theory amp Applications vol 4 no 5 pp 735ndash7452010
[41] F Amato M Ariola and C Cosentino ldquoFinite-time control ofdiscrete-time linear systems analysis and design conditionsrdquoAutomatica vol 46 no 5 pp 919ndash924 2010
[42] X Lin H Du and S Li ldquoFinite-time boundedness and 1198712-
gain analysis for switched delay systems with norm-boundeddisturbancerdquo Applied Mathematics and Computation vol 217no 12 pp 5982ndash5993 2011
[43] J Lin S Fei and Z Gao ldquoStabilization of discrete-time switchedsingular time-delay systems under asynchronous switchingrdquoJournal of the Franklin Institute vol 349 no 5 pp 1808ndash18272012
[44] H Song L Yu D Zhang and W-A Zhang ldquoFinite-time 119867infin
control for a class of discrete-time switched time-delay systemswith quantized feedbackrdquoCommunications in Nonlinear Scienceand Numerical Simulation vol 17 no 12 pp 4802ndash4814 2012
[45] Z Zuo H Li Y Liu and Y Wang ldquoOn finite-time stochasticstability and stabilization ofMarkovian jump systems subject topartial information on transition probabilitiesrdquoCircuits Systemsand Signal Processing 2012
[46] H Liu Y Shen and X Zhao ldquoDelay-dependent observer-based119867
infinfinite-time control for switched systems with time-varying
delayrdquoNonlinear Analysis Hybrid Systems vol 6 no 3 pp 885ndash898 2012
[47] Y Zhang C Liu and X Mu ldquoRobust finite-time119867infincontrol of
singular stochastic systems via static output feedbackrdquo AppliedMathematics and Computation vol 218 no 9 pp 5629ndash56402012
Journal of Applied Mathematics 13
[48] J Cheng H Zhu S Zhong Y Zhang and Y Zeng ldquoFinite-timeStabilization of 119867
infinfiltering for switched stochastic systemsrdquo
Circuits Systems and Signal Processing 2012[49] J Cheng H Zhu S Zhong and Y Zhang ldquoFinite-time bound-
ness of 119867infin
filtering for switching discrete-time systemsrdquo In-ternational Journal of Control Automation and Systems vol 10pp 1129ndash1135 2012
[50] Z Xiang C Qiao and M S Mahmoud ldquoFinite-time analysisand119867
infincontrol for switched stochastic systemsrdquo Journal of the
Franklin Institute vol 349 no 3 pp 915ndash927 2012
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
6 Journal of Applied Mathematics
+ sum
119899isinM119898 = 119899
120587119898119899119876
2119894minus 119876)
times 120578 (119904) 119889119904
(28)
Since
0 le 120591119898(119905) le ℎ
119898 (29)
we define
119903 (120591119898) =
minus (1 minus 120583119898) 119890
120575ℎ119898 if 120583
119898gt 1
minus (1 minus 120583119898) if 120583
119898le 1
(30)
Then
pound1198812(120578
119905 119894 119898) le 119890
120575119905
120578⊺
(119905) (119890120575ℎ
1198761119894+ 119890
120575ℎ
1198762119894+ ℎ119876) 120578 (119905)
minus 119890120575119905
120578⊺
(119905 minus ℎ)1198761119894120578 (119905 minus ℎ)
+ 119890120575119905
120578⊺
(119905 minus 120591119898(119905)) 119903 (120591
119898) 119876
2119894120578 (119905 minus 120591
119898(119905))
+ int
119905
119905minusℎ
119890120590(119904+ℎ)
120578⊺
(119904)
times ( sum
119895isinN119895 = 119894
120582119894119895(119876
1119895+ 119876
2119895)
+ sum
119899isinM119898 = 119899
120587119898119899119876
2119894minus 119876)
times 120578 (119904) 119889119904
(31)
Similar to the previous process we can obtain
pound1198813(120578
119905 119894 119898) le int
0
minusℎ
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904)
times ( sum
119899isinM119898 = 119899
120587119898119899119883
119894119899
+ sum
119895isinN119894 = 119895
120582119894119895119883
119895119898minus 119883)
times 120578 (119904) 119889119904 119889120579
+ 119890120575119905
120578⊺
(119905) 119883119894119898120578 (119905) int
0
minusℎ
119890minus120575120592
119889120592
minus 119890120575119905
int
119905
119905minusℎ
120578⊺
(119904)119883119894119898120578 (119904) 119889119904
+ 119890120575119905
120578⊺
(119905) 119883120578 (119905) int
0
minusℎ
int
0
120592
119890minus120575120592
119889120579119889120592
(32)
By using Lemma 5 it yields that
minusint
119905
119905minusℎ
120578⊺
(119904)119883119894119898120578 (119904) 119889119904 = minusint
119905
119905minus120591119898(119905)
120578⊺
(119904)119883119894119898120578 (119904) 119889119904
minus int
119905minus120591119898(119905)
119905minusℎ
120578⊺
(119904)119883119894119898120578 (119904) 119889119904
le minus120591119898(119905) 119880
⊺
1119883
1198941198981198801
minus (ℎ minus 120591119898(119905)) 119880
⊺
2119883
1198941198981198802
(33)
where
1198801=
1
120591119898(119905)
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904
1198802=
1
ℎ minus 120591119898(119905)
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904
lim120591119898(119905)rarr0
1
120591119898(119905)
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904 = 120578 (119905)
lim120591119898(119905)rarrℎ
1
ℎ minus 120591119898(119905)
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904 = 120578 (119905 minus ℎ)
(34)
From the Newton-Leibniz formula the following equa-tion is true for any matrices119872
119894119898 119873
119894119898 and 119881
119894119898with appro-
priate dimensions
(2120591119898(119905) 119880
⊺
1119872
119894119898+ 2 (ℎ minus 120591
119898(119905)) 119880
⊺
2119873
119894119898+ 2 120578
⊺
(119905) 119881119894119898)
times [minus 120578 (119905) + 119860119894120578 (119905) + 119861
120591119894120578 (119905 minus 120591
119898(119905)) + 119863
119894120596 (119905)] = 0
(35)
pound1198814(120578
119905 119894 119898)
le int
0
minusℎ
int
119905
119905+120579
119890120575(119904minus120579)
120578⊺
(119904)
times ( sum
119899isinM119898 = 119899
120587119898119899119884119894119899
+ sum
119895isinN119894 = 119895
120582119894119895119884119895119898minus 119884)
times 120578 (119904) 119889119904 119889120579
+ 119890120575119905
120578⊺
(119905) 119884119894119898120578 (119905) int
0
minusℎ
119890minus120575120592
119889120592
minus 119890120575119905
int
119905
119905minusℎ
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
+ 119890120575119905
120578⊺
(119905) 119884 120578 (119905) int
0
minusℎ
int
0
120592
119890minus120575120592
119889120579 119889120592
(36)
Journal of Applied Mathematics 7
From Lemma 5 it yields that
minus int
119905
119905minusℎ
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
= minusint
119905
119905minus120591119898(119905)
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
minus int
119905minus120591119898(119905)
119905minusℎ
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
le minus
ℎ
120591119898(119905)
[int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904]
⊺
119884119894119898
ℎ
[int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904]
minus
ℎ
ℎ minus 120591119898(119905)
[int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904]
⊺
119884119894119898
ℎ
[int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904]
le minus
[
[
[
[
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904
]
]
]
]
⊺
[
[
[
119884119894119898
ℎ
119878119894119898
lowast
119884119894119898
ℎ
]
]
]
[
[
[
[
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904
]
]
]
]
(37)
From (25)ndash(37) we can eventually obtain
pound119881 (120578119905 119903
119905 119904
119905) minus 120575120596
⊺
(119905)119867120596 (119905) le 119890120575119905
120585⊺
(119905) Ξ119894119898120585 (119905) (38)
where
120585⊺
(119905) = [120578⊺
(119905) 120578⊺
(119905 minus 120591119898(119905)) 120578
⊺
(119905 minus ℎ) 120578⊺
(119905) 119880⊺
1 119880
⊺
2 120596
⊺
(119905)]
Ξ119894119898=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860⊺
119894119881
⊺
119894119898120591119898(119905) 119860
⊺
119894119872
⊺
119894119898(ℎ minus 120591
119898(119905)) 119860
⊺
119894119873
⊺
119894119898119875119894119898119863
119894
lowast Ξ22119894119898
119878119894119898minus
119884119894119898
ℎ
119860⊺
120591119894119881
⊺
119894119898120591119898(119905) 119860
⊺
120591119894119872
⊺
119894119898(ℎ minus 120591
119898(119905)) 119860
⊺
120591119894119873
⊺
1198941198980
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minus120591119898(119905)119872
⊺
119894119898minus (ℎ minus 120591
119898(119905))119873
⊺
119894119898119881119894119898119863
119894
lowast lowast lowast lowast minus120591119898(119905) 119883
1198941198980 120591
119898(119905)119872
119894119898119863
119894
lowast lowast lowast lowast lowast minus (ℎ minus 120591119898(119905))119883
119894119898(ℎ minus 120591
119898(119905))119873
119894119898119863
119894
lowast lowast lowast lowast lowast lowast minus120575119867
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
(39)
The LMIs (20) and (21) lead to 120591119898(119905) rarr ℎ and 120591
119898(119905) rarr
0 respectively It is easy to see that Ξ1119894119898
results from Ξ119894119898|
120591119898(119905) = ℎ and Ξ
119894119898|120591119898(119905) = 0 Thus we obtain
E pound119881 (120578119905 119903
119905 119904
119905) le 120578E [119881 (120578
119905 119903
119905 119904
119905)] + 120575120596
⊺
(119905)119867120596 (119905)
(40)
Multiplying the aforementioned inequality by 119890minus120578119905 we canget
E pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] le 119890
minus120578119905
120575120596⊺
(119905)119867120596 (119905) (41)
By integrating the aforementioned inequality between 0and 119905 it follows that
119890minus120578119905
E [119881 (120578119905 119903
119905 119904
119905)] minus E [119881 (120578
0 119903
0 119904
0)]
le 120575int
119905
0
119890minus120578119904
120596⊺
(119904)119867120596 (119904) 119889119904
(42)
Denote that 119894119898
= 119877minus(12)
119875119894119898119877minus(12) 119876
119904119894=
119877minus(12)
119876119904119894119877minus(12)
(119904 = 1 2) 119876 = 119877minus(12)
119876119877minus(12)
119883119894119898
= 119877minus(12)
119883119894119898119877minus(12) 119883 = 119877
minus(12)
119883119877minus(12)
119894119898
= 119877minus(12)
119884119894119898119877minus(12) and = 119877
minus(12)
119884119877minus(12) it yields
that
E [119881 (1205780 119903
0 119904
0)]
le max119894isinN119898isinM
120582max (119894119898) 120578⊺
(0) 119877120578 (0)
+ (max119894isinN
120582max (1198761119894) +max
119894isinN120582max (1198762119894
)) 119890120575ℎ
times int
0
minusℎ
119890120575119904
120578⊺
(119904) 119877120578 (119904) 119889119904
+ 119890120575ℎ
120582max (119876)int0
minusℎ
int
0
120579
119890120575119904
120578⊺
(119904) 119877120578 (119904) 119889119904
+ 119890120575ℎ max
119894isinN119898isinM120582max (119883119894119898
)
times int
0
minusℎ
int
0
120579
119890minus120575120579
120578⊺
(119904) 119877120578 (119904) 119889119904 119889120579
+ 119890120575ℎ
120582max (119883)
times int
0
minusℎ
int
0
120579
int
0
120592
119890minus120575120579
120578⊺
(119904) 119877120578 (119904) 119889119904 119889120579 119889120592
+ 119890120575ℎ max
119894isinN119898isinM120582max (119894119898)
8 Journal of Applied Mathematics
times int
0
minusℎ
int
0
120579
119890minus120575120579
120578⊺
(119904) 119877 120578 (119904) 119889119904 119889120579
+ 119890120575ℎ
120582max ()
times int
0
minusℎ
int
0
120579
int
0
120592
119890minus120575120579
120578⊺
(119904) 119877 120578 (119904) 119889119904 119889120579 119889120592
le max119894isinN119898isinM
120582max (119894119898)
+ ℎ119890120575ℎ
(max119894isinN
120582max (1198761119894+
119894isinNmax120582max (1198762119894
)))
+ ℎ2
119890120575ℎ
( max119894isinN119898isinM
120582max (119883119894119898) + 120582max (119876)
+ max119894isinN119898isinM
120582max (119894119898))
+
1
2
ℎ3
119890120575ℎ
(120582max (119883) + 120582max ())
times supminusℎle119904le0
120578⊺
(119904) 119877120578 (119904) 120578⊺
(119904) 119877 120578 (119904)
= 1198881Λ
(43)
For given 120578 gt 0 and 0 le 119905 le 119879 we have
E [119881 (120578119905 119903
119905 119904
119905)] le E [119890
120578119905
119881 (1205780 119903
0 119904
0)]
+ 119890120578119905
120575int
119905
0
119890minus120578119904
120596⊺
(119904)119867120596 (119904) 119889119904
le 119890120578119879
1198881Λ + 119889120575119890
120578119879
120582max (119867)int119879
0
119890minus120578119904
119889119904
le 119890120578119879
1198881Λ + 119889120575120582
10
1
120578
(1 minus 119890minus120578119879
)
(44)
On the other hand it follows from (25) that
E [119881 (120578119905 119903
119905 119904
119905)] ge E [120578
⊺
(119905) 119890120582119905
119875119894119898120578 (119905)]
ge max119894isinN
120582min (119875119894119898)E [120578⊺
(119905) 119877120578 (119905)]
= 1205821E [120578
⊺
(119905) 119877120578 (119905)]
(45)
It can be derived from (43)ndash(45) that
E [120578⊺
(119905) 119877120578 (119905)] le
119890120578119879
1205821
1198881Λ + 119889120575120582
10
1
120578
(1 minus 119890minus120578119879
) (46)
From (22) and (46) we have
E [120578⊺
(119905) 119877120578 (119905)] lt 1198882 (47)
Then the system is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879)
Remark 8 In this paper 120591119898(119905) and 120591
119898(119905) may have different
upper bounds in various delay intervals satisfying (7) respec-tively However in previous work such as [20 21] 120591
119898(119905) and
120591119898(119905) are enlarged to 120591
119898(119905) le ℎ = maxℎ
119894 119898 isin M and
120591119898(119905) le 120583 = max120583
119898 119898 isin M respectively which may lead
to conservativeness inevitably However the previouse casecan be taken fully into account by employing the Lyapunov-Krasovskii functional (25)
Remark 9 When dealing with termminusint
119905
119905minusℎ
120578⊺
(119904)119884119894119898120578(119904)119889119904 the
convex combination is not employed Lemma 5 is used in thispaper and then the free-weighting matrices-dependent nulladd items are necessary to be introduced in our proof whichlead to the decrease of the number of LMIs and LMIs scalardecision variables
Remark 10 The feature of this paper is the way to dealwith the integral term Many researchers have enlarged thederivative of the Lyapunov functional in order to deal with theintegral term in mathematical operations In this paper wepropose a novel delay-dependent sufficient criterion whichensures that the Markovian jump system with different modesystems is finite-time stable
Remark 11 It should be pointed out that the novelty of theLyapunov functional (25) lies in distinct Lyapunov matrices(119875
119894119898 119883
119894119898 119884
119894119898) which is chosen for different system modes
119894 (119894 = 1 2 119873) and119898 (119898 = 1 2 119872)
Theorem 12 System (9) is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879) if there exist matrices 119875
119894gt 0 119876
119897119894gt 0 (119897 = 1 2)
119876 gt 0 119883119894119898gt 0 119883 gt 0 119884
119894119898gt 0 119884 gt 0 119878
119894119898 scalars 119888
1lt 119888
2
119879 gt 0 120582119904gt 0 (119904 = 1 2 9) 120578 gt 0 120574 gt 0 and Λ gt 0 such
that for all 119894 119895 isinN and119898 119899 isinM and (16)ndash(19) the followinginequalities hold
Σ1119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119872⊺
119894119898119875119894119898119863119894
119862
⊺
119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119872⊺
1198941198980 119864
⊺
119894
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119872⊺
119894119898119881119894119898119863119894
0
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119872119894119898119863119894
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(48)
Σ2119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119873⊺
119894119898119875119894119898119863119894
119862
⊺
119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119873⊺
1198941198980 119864
⊺
119894
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119873⊺
119894119898119881119894119898119863119894
0
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119873119894119898119863119894
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(49)
1198881Λ + 119889120575120574
21
120578
(1 minus 119890minus120578119879
) lt 1205821119890minus120578119879
1198882 (50)
Journal of Applied Mathematics 9
Proof We now consider the 119867infin
performance of system (9)Select the same Lyapunov-Krasovskii functional as Theo-rem 7 it yields that
pound119881 (120578119905 119903
119905 119904
119905) + 119890
⊺
(119905) 119890 (119905) minus 1205742
120596⊺
(119905) 120596 (119905) le 120585⊺
(119905) Σ119897119894119898120585 (119905)
(119897 = 1 2)
(51)
It follows from (49)-(50) that
E pound119881 (120578119905 119903
119905 119904
119905)
le E [120578119881 (120578119905 119903
119905 119904
119905)] + 120574
2
120596⊺
(119905) 120596 (119905) minus E [119890⊺
(119905) 119890 (119905)]
(52)
Multiplying the aforementioned inequality by 119890minus120578119905 onehas
E pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] le 119890
minus120578119905
[1205742
120596⊺
(119905) 120596 (119905) minus 119890⊺
(119905) 119890 (119905)]
(53)
In zero initial condition and E[119881(120578119905 119903
119905 119904
119905)] gt 0 by inte-
grating the aforementioned inequality between 0 and 119879 wecan get
int
119879
0
119890minus120578119905
[1205742
120596⊺
(119905) 120596 (119905) minus 119890⊺
(119905) 119890 (119905)] 119889119905
le Eint119879
0
pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] 119889119905 le 119881 (120578
0 119903
0 119904
0) = 0
(54)
Using Dynkins formula it results that
E [int119879
0
119890minus120578120592
119890⊺
(120592) 119890 (120592) 119889120592] le 1205742
E [int119879
0
119890minus120578120592
120596⊺
(120592) 120596 (120592) 119889120592]
(55)
Then it yields that
E [int119879
0
119890⊺
(120592) 119890 (120592) 119889120592] le 1205742
119890120578119879
E [int119879
0
120596⊺
(120592) 120596 (120592) 119889120592] (56)
Thus it is concluded by Definition 3 that system (9) is fi-nite-time bounded with an 119867
infinperformance 120574 This com-
pletes the proof
Remark 13 From the proof process of Theorems 7 and 12it is easy to see that neither bounding technique for cross-terms nor model transformation is involved In other wordsthe obtained result is expected to be less conservative
Remark 14 The Lyapunov asymptotic stability and finite-time stability of a class of system are independent conceptsA Lyapunov asymptotic stability system may not be finite-time stability Moreover finite-time stability system may alsonot be Lyapunov asymptotic stabilityThere exist some resultson Lyapunov stability while finite-time stability also needsour full investigation which was neglected by most previouswork
4 Finite-Time 119867infin
Filtering
Theorem 15 System (9) is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879) if there exist matrices 119875
119894119898gt 0119872
119894119898 119873
119894119898 119881
119894119898
119876119897119894gt 0 (119897 = 1 2) 119876 gt 0 119883
119894119898gt 0 119883 gt 0 119884
119894119898gt 0 119884 gt 0
119860119891119894119898
119861119891119894119898
119862119891119894119898
119872119860119894119898
119872119861119894119898
119873119860119894119898
119873119861119894119898
119881119860119894119898
119881119861119894119898
119878119894119898
scalars 1198881lt 119888
2 119879 gt 0 120590
119904gt 0 (119904 = 1 2 9) 120575 gt 0 120578 gt 0
and Λ gt 0 such that for all 119894 119895 isin N and 119898 119899 isin M thefollowing inequalities hold
119875119894119898= [
1198751119894119898
1198752119894119898
1198752119894119898
1198752119894119898
] gt 0 119872119894119898= [
1198721119894119898
1198722119894119898
1198722119894119898
1198722119894119898
] (57)
119873119894119898= [
1198731119894119898
1198732119894119898
1198732119894119898
1198732119894119898
] 119881119894119898= [
1198811119894119898
1198812119894119898
1198812119894119898
1198812119894119898
] (58)
Γ1119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Γ11119894119898
Γ12119894119898
minus119878119894119898
Γ14119894119898
Γ15119894119898
Γ16119894119898
Γ17119894119898
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
Γ24119894119898
Γ25119894119898
0 [119862⊺
119911120591119894
0]
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119872⊺
119894119898Γ46119894119898
0
lowast lowast lowast lowast minusℎ119883119894119898
Γ56119894119898
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(59)
Γ2119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Γ11119894119898
Γ12119894119898
minus119878119894119898
Γ14119894119898
Γ1015840
15119894119898Γ16119894119898
Γ17119894119898
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
Γ24119894119898
Γ1015840
251198941198980 [
119862⊺
119911120591119894
0]
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119873⊺
119894119898Γ1015840
461198941198980
lowast lowast lowast lowast minusℎ119883119894119898
Γ1015840
561198941198980
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(60)where
Γ11119894119898
= sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898+ 119890
120575ℎ
1198761119894+ 119890
120575ℎ
1198762119894+ ℎ119876 +
119890120575ℎ
minus 1
120575
119883119894119898+
119890120575ℎ
minus 120575ℎ119890120575ℎ
minus 1
1205752
119883
+
119884119894119898
ℎ
+ [
1205751198751119894119898
+ 1198751119894119898119860
119894+ 119860
⊺
1198941198751119894119898
+ 119861119891119894119898119862119910119894+ 119862
⊺
119910119894119861
⊺
119891119894119898120575119875
2119894119898+ 119860
119891119894119898+ 119860
⊺
119891119894119898
1205751198752119894119898
+ 1198752119894119898119860
119894+ 119860
⊺
1198941198752119894119898
+ 119861119891119894119898119862119910119894+ 119862
⊺
119910119894119861
⊺
119891119894119898120575119875
2119894119898+ 119860
119891119894119898+ 119860
⊺
119891119894119898
]
(61)
10 Journal of Applied Mathematics
Γ12119894119898
= minus
119884119894119898
ℎ
+ 119878119894119898+ [
1198751119894119898119860
120591119894+ 119861
119891119894119898119862119910120591119894
1198752119894119898119860
120591119894+ 119861
119891119894119898119862119910120591119894
]
Γ14119894119898
= [
119860⊺
119894119881
⊺
1119894119898+ 119862
⊺
119910119894119881
⊺
119861119894119898119881
⊺
119860119894119898
119860⊺
119894119881
⊺
2119894119898+ 119862
⊺
119910119894119881
⊺
119861119894119898119881
⊺
119860119894119898
]
Γ15119894119898
= [
ℎ119860⊺
119894119872
⊺
1119894119898+ ℎ119862
⊺
119910119894119872
⊺
119861119894119898ℎ119872
⊺
119860119894119898
ℎ119860⊺
119894119872
⊺
2119894119898+ ℎ119862
⊺
119910119894119872
⊺
119861119894119898ℎ119872
⊺
119860119894119898
]
Γ1015840
15119894119898= [
ℎ119860⊺
119894119873
⊺
1119894119898+ ℎ119862
⊺
119910119894119873
⊺
119861119894119898ℎ119873
⊺
119860119894119898
ℎ119860⊺
119894119873
⊺
2119894119898+ ℎ119862
⊺
119910119894119873
⊺
119861119894119898ℎ119873
⊺
119860119894119898
]
Γ16119894119898
= [
1198751119894119898119863
119894+ 119861
119891119894119898119863
119910119894
1198752119894119898119863
119894+ 119861
119891119894119898119863
119910119894
]
Γ17119894119898
= [
119862⊺
119911119894
minus119862119891119894119898
]
Γ24119894119898
= [
119860⊺
120591119894119881
⊺
1119894119898+ 119862
⊺
119910120591119894119881
⊺
119861119894119898
119860⊺
120591119894119881
⊺
2119894119898+ 119862
⊺
119910120591119894119881
⊺
119861119894119898
]
Γ25119894119898
= [
ℎ119860⊺
120591119894119872
⊺
1119894119898+ ℎ119862
⊺
119910120591119894119872
⊺
119861119894119898
ℎ119860⊺
120591119894119872
⊺
2119894119898+ ℎ119862
⊺
119910120591119894119872
⊺
119861119894119898
]
Γ1015840
25119894119898= [
ℎ119860⊺
120591119894119873
⊺
1119894119898+ ℎ119862
⊺
119910120591119894119873
⊺
119861119894119898
ℎ119860⊺
120591119894119873
⊺
2119894119898+ ℎ119862
⊺
119910120591119894119873
⊺
119861119894119898
]
Γ46119894119898
= [
1198811119894119898119863
119894+ 119881
119861119894119898119863
119910119894
1198812119894119898119863
119894+ 119881
119861119894119898119863
119910119894
]
Γ56119894119898
= [
ℎ1198721119894119898119863
119894+ ℎ119872
119861119894119898119863
119910119894
ℎ1198722119894119898119863
119894+ ℎ119872
119861119894119898119863
119910119894
]
Γ1015840
56119894119898= [
ℎ1198731119894119898119863
119894+ ℎ119873
119861119894119898119863
119910119894
ℎ1198732119894119898119863
119894+ ℎ119873
119861119894119898119863
119910119894
]
(62)
Then a desired filter can be chosen with parameters as
119860119891119894119898
= 119875minus1
2119894119898119860
119891119894119898 119861
119891119894119898= 119875
minus1
2119894119898119861119891119894119898
119862119891119894119898
= 119862119891119894119898
(63)
Proof We denote that
119875119894119898= [
1198751119894119898
1198752119894119898
1198752119894119898
1198752119894119898
] (64)
The term 119875119894119898119860
119894can be rewritten as
119875119894119898119860
119894= [
1198751119894119898119860
119894+ 119875
21198941198981198611198911198941198981198621199101198941198752119894119898119860
119891119894119898
1198752119894119898119860
119894+ 119875
21198941198981198611198911198941198981198621199101198941198752119894119898119860
119891119894119898
] (65)
Similarly we have
119875119894119898119861119894= [
1198751119894119898119860
120591119894+ 119875
2119894119898119861119891119894119898119862119910120591119894
1198752119894119898119860
120591119894+ 119875
2119894119898119861119891119894119898119862119910120591119894
]
119875119894119898119863
119894= [
1198751119894119898119863
119894+ 119875
2119894119898119861119891119894119898119863
119910119894
1198752119894119898119863
119894+ 119875
2119894119898119861119891119894119898119863
119910119894
]
(66)
Define 119860119891119894119898
= 1198752119894119898119860
119891119894119898 119861
119891119894119898= 119875
2119894119898119861119891119894119898
119862119891119894119898
= 119862119891119894119898
119872
119860119894119898= 119872
2119894119898119860
119891119894119898119872
119861119894119898= 119872
2119894119898119861119891119894119898
119873119860119894119898
= 1198732119894119898119860
119891119894119898
119873119861119894119898
= 1198732119894119898119861119891119894119898
119881119860119894119898
= 1198812119894119898119860
119891119894119898 and 119881
119861119894119898= 119881
2119894119898119861119891119894119898
Therefore if (59) and (60) hold system (9) is finite-timebounded with a prescribed 119867
infinperformance index 120574 The
proof is completed
Remark 16 In many actual applications the minimum valueof 1205742min is of interest In Theorem 12 with a fixed 120582 120574min canbe obtained through the following optimization procedure
min 1205742
st (48)ndash(50) (67)
In Theorem 15 as for finite-time stability and bounded-ness once the state bound 119888
2is not ascertained theminimum
value 1198882min is of interest With a fixed 120582 define 120582
1= 1 then
the following optimization problem can be formulated to getminimum value 119888
2min
min 1205891205742
+ (1 minus 120589) 1198882
st (59)-(60) and (50)
(68)
where 120589 is weighted factor and 120589 isin [0 1]
5 Illustrative Example
Example 17 Consider that the Markovian jump system andthe delay mode switching are governed by a Markov processwith the following transition rates
Λ = [
minus07 07
09 minus09] Π = [
minus1 1
12 minus12] (69)
as well as with the following parameters
1198601= [
minus35 086
minus064 minus325] 119860
2= [
minus25 034
14 minus002]
1198601205911= [
minus08 minus13
minus07 minus22] 119860
1205912= [
minus28 05
minus08 minus14]
1198631= 119863
2= [
01
02] 119862
1199101= [04 07]
1198621199102= [05 08] 119862
1199101205911= [01 02]
1198621199101205912
= [01 01]
1198631199101= 119863
1199102= 01 119862
1199111= [01 005]
1198621199112= [02 009] 119862
1199111205911= [03 06]
1198621199111205912
= [04 08] 1198631199111= 119863
1199112= 005
(70)
Journal of Applied Mathematics 11
Then we choose 119877 = 119868 119879 = 2 1198881= 1 and 119889 = 001
throughTheorem 15 it yields that themode-dependent filtersare as follows
11986011989111
= [
minus61254 15565
minus02347 minus03763]
11986011989112
= [
minus62682 14338
minus03552 minus05327]
11986011989121
= [
minus89214 24223
minus15476 minus06234]
11986011989122
= [
minus84638 28237
minus13545 minus04322]
11986111989111
= [
51465 minus28516
minus91216 95171]
11986111989112
= [
54203 minus23121
minus93156 96332]
11986111989121
= [
46193 minus11984
minus162397 163006]
11986111989122
= [
46512 minus10311
minus161132 162312]
11986211989111
= [00294 minus00397]
11986211989112
= [00271 minus003368]
11986211989121
= [01163 minus01432]
11986211989122
= [01342 minus03672]
(71)
This paper deals with the finite-time filter design problemfor a class of Markovian jump systems particularly two dif-ferentMarkov processes are considered formodeling the ran-domness of system matrix and the state delay Then throughthe numerical example we can see that results in this paperare feasible which further verified the correctness of our the-ory Therefore the paper shorten this gap
6 Conclusions
In this paper we have examined the problems of finite-time119867
infinfiltering for a class of Markovian jump systems with dif-
ferent system modes Based on a novel approach a sufficientcondition is derived such that the closed-loop Markovianjump system is finite-time bounded and satisfies a prescribedlevel of119867
infindisturbance attenuation in a finite time interval
Finally a numerical example is also given to illustrate theeffectiveness of the proposed design approach It should benoted that one of future research topics would be to investi-gate the problems of fault detection and fault tolerant controlfor time-varying Markovian jump systems with incompleteinformation over a finite-time horizon
Acknowledgments
The authors would like to thank the associate editor and theanonymous reviewers for their detailed comments and sug-gestions This work was supported by the Fund of SichuanProvincial Key Laboratory of Signal and Information Pro-cessing Xihua University (SZJJ2009-002 and SGXZD0101-10-1) and National Basic Research Program of China(2010CB732501)
References
[1] N N Krasovskiı and E A Lidskiı ldquoAnalytical design ofcontrollers in systems with random attributesmdashI Statementof the problem method of solvingrdquo Automation and RemoteControl vol 22 pp 1021ndash1025 1961
[2] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[3] L Wu X Su and P Shi ldquoSliding mode control with bounded1198712gain performance of Markovian jump singular time-delay
systemsrdquo Automatica vol 48 no 8 pp 1929ndash1933 2012[4] H Gao Z Fei J Lam and B Du ldquoFurther results on expo-
nential estimates of Markovian jump systems with mode-de-pendent time-varying delaysrdquo IEEE Transactions on AutomaticControl vol 56 no 1 pp 223ndash229 2011
[5] Z Wu H Su and J Chu ldquo119867infinfiltering for singular Markovian
jump systems with time delayrdquo International Journal of Robustand Nonlinear Control vol 20 no 8 pp 939ndash957 2010
[6] P Tino M Cernansky and L Benuskova ldquoMarkovian architec-tural bias of recurrent neural networksrdquo IEEE Transactions onNeural Networks vol 15 no 1 pp 6ndash15 2004
[7] H Dong Z Wang D W C Ho and H Gao ldquoRobust 119867infin
filtering for Markovian jump systems with randomly occurringnonlinearities and sensor saturation the finite-horizon caserdquoIEEE Transactions on Signal Processing vol 59 no 7 pp 3048ndash3057 2011
[8] Q Ma S Xu and Y Zou ldquoStability and synchronization forMarkovian jump neural networks with partly unknown tran-sition probabilitiesrdquo Neurocomputing vol 74 pp 3404ndash33412011
[9] H Zhao S Xu and Y Zou ldquoRobust119867infinfiltering for uncertain
Markovian jump systems with mode-dependent distributeddelaysrdquo International Journal of Adaptive Control and SignalProcessing vol 24 no 1 pp 83ndash94 2010
[10] Y Zhang S Xu and B Zhang ldquoDiscrete-time fuzzy Markovianjump systems with time-varying delaysrdquo IEEE Transactions onFuzzy Systems vol 17 no 2 pp 411ndash420 2009
[11] Z Wu H Su and J Chu ldquoState estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 no 10ndash12 pp 2247ndash2254 2010
[12] Q Zhu and J Cao ldquoRobust exponential stability of markovianjump impulsive stochastic Cohen-Grossberg neural networkswithmixed time delaysrdquo IEEE Transactions onNeural Networksvol 21 no 8 pp 1314ndash1325 2010
[13] H Huang G Feng and X Chen ldquoStability and stabilizationof Markovian jump systems with time delay via new Lyapunovfunctionalsrdquo IEEE Transactions on Circuits and Systems I vol59 no 10 pp 2413ndash2421 2012
12 Journal of Applied Mathematics
[14] Q Zhu and J Cao ldquoStability of Markovian jump neural net-works with impulse control and time varying delaysrdquoNonlinearAnalysis Real World Applications vol 13 no 5 pp 2259ndash22702012
[15] M Sun J Lam S Xu andY Zou ldquoRobust exponential stabiliza-tion for Markovian jump systems with mode-dependent inputdelayrdquo Automatica vol 43 no 10 pp 1799ndash1807 2007
[16] M Mohammadian A Abolmasoumi and H Momeni ldquo119867infin
mode-independent filter design for Markovian jump geneticregulatory networks with time-varying delaysrdquo Neurocomput-ing vol 87 pp 10ndash18 2012
[17] S Boyd L El Ghaoui E Feron and V Balakrishnan LinearMatrix Inequalities in SystemandControlTheory vol 15 of SIAMStudies in Applied Mathematics SIAM Philadelphia Pa USA1994
[18] S Mou H GaoW Qiang and K Chen ldquoNew delay-dependentexponential stability for neural networks with time delayrdquo IEEETransactions on Systems Man and Cybernetics Part B vol 38no 2 pp 571ndash576 2008
[19] O M Kwon J Park S Lee and E Cha ldquoAnalysis on delay-dependent stability for neural networks with time-varyingdelaysrdquo Neurocomputing vol 103 pp 114ndash120 2013
[20] L Hu H Gao andW X Zheng ldquoNovel stability of cellular neu-ral networkswith interval time-varying delayrdquoNeuralNetworksvol 21 no 10 pp 1458ndash1463 2008
[21] H Zhang Z Liu G B Huang and ZWang ldquoNovel weighting-delay-based stability criteria for recurrent neural networks withtime-varying delayrdquo IEEE Transactions on Neural Networks vol21 no 1 pp 91ndash106 2010
[22] Z Zuo C Yang and YWang ldquoA newmethod for stability anal-ysis of recurrent neural networks with interval time-varyingdelayrdquo IEEE Transactions on Neural Networks vol 21 no 2 pp339ndash344 2010
[23] P Park JWKo andC Jeong ldquoReciprocally convex approach tostability of systems with time-varying delaysrdquo Automatica vol47 no 1 pp 235ndash238 2011
[24] D Zhang L Yu and W Zhang ldquoDelay-dependent fault detec-tion for switched linear systems with time-varying delays-theaverage dwell time approachrdquo Signal Processing vol 91 pp 832ndash840 2011
[25] Z Fei H Gao and P Shi ldquoNew results on stabilization ofMark-ovian jump systems with time delayrdquoAutomatica vol 45 no 10pp 2300ndash2306 2009
[26] H R Karimi ldquoRobust delay-dependent 119867infin
control of uncer-tain time-delay systems with mixed neutral discrete anddistributed time-delays and Markovian switching parametersrdquoIEEE Transactions on Circuits and Systems I vol 58 no 8 pp1910ndash1923 2011
[27] M Luo G Liu and S Zhong ldquoRobust fault detection ofMarkovian jump systems with different systemmodesrdquoAppliedMathematical Modelling vol 37 no 7 pp 5001ndash5012 2013
[28] B Zhang andY Li ldquoExponential1198712minus119871
infinfiltering for distributed
delay systems with Markovian jumping parametersrdquo SignalProcessing vol 93 no 1 pp 206ndash216 2013
[29] A Abolmasoumi and H Momeni ldquoRobust observer-based119867infin
control of a Markovian jump system with different delay andsystem modesrdquo International Journal of Control Automationand Systems vol 9 no 4 pp 768ndash776 2011
[30] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[31] S Ma and E K Boukas ldquoRobust 119867infin
filtering for uncertaindiscrete Markov jump singular systems with mode-dependenttime delayrdquo IET Control Theory amp Applications vol 3 no 3 pp351ndash361 2009
[32] Y ChenW Bi andW Li ldquoStability analysis for neural networkswith time-varying delay a more general delay decompositionapproachrdquo Neurocomputing vol 73 no 4ndash6 pp 853ndash857 2010
[33] Z Shu J Lam and S Xu ldquoRobust stabilization of Markoviandelay systems with delay-dependent exponential estimatesrdquoAutomatica vol 42 no 11 pp 2001ndash2008 2006
[34] X Huang W Lin and B Yang ldquoGlobal finite-time stabilizationof a class of uncertain nonlinear systemsrdquo Automatica vol 41no 5 pp 881ndash888 2005
[35] C Qian and J Li ldquoGlobal finite-time stabilization by outputfeedback for planar systems without observable linearizationrdquoIEEE Transactions on Automatic Control vol 50 no 6 pp 885ndash890 2005
[36] F Amato R Ambrosino C Cosentino and G De TommasildquoInput-output finite time stabilization of linear systemsrdquo Auto-matica vol 46 no 9 pp 1558ndash1562 2010
[37] Y Hong ldquoFinite-time stabilization and stabilizability of a classof controllable systemsrdquo Systems amp Control Letters vol 46 no4 pp 231ndash236 2002
[38] F Amato and M Ariola ldquoFinite-time control of discrete-timelinear systemsrdquo IEEETransactions onAutomatic Control vol 50no 5 pp 724ndash729 2005
[39] Z Zuo Y Liu Y Wang and H Li ldquoFinite-time stochasticstability and stabilisation of linear discrete-time Markovianjump systems with partly unknown transition probabilitiesrdquoIET Control Theory amp Applications vol 6 no 10 pp 1522ndash15262012
[40] X Luan F Liu and P Shi ldquoFinite-time filtering for non-linearstochastic systems with partially known transition jump ratesrdquoIET Control Theory amp Applications vol 4 no 5 pp 735ndash7452010
[41] F Amato M Ariola and C Cosentino ldquoFinite-time control ofdiscrete-time linear systems analysis and design conditionsrdquoAutomatica vol 46 no 5 pp 919ndash924 2010
[42] X Lin H Du and S Li ldquoFinite-time boundedness and 1198712-
gain analysis for switched delay systems with norm-boundeddisturbancerdquo Applied Mathematics and Computation vol 217no 12 pp 5982ndash5993 2011
[43] J Lin S Fei and Z Gao ldquoStabilization of discrete-time switchedsingular time-delay systems under asynchronous switchingrdquoJournal of the Franklin Institute vol 349 no 5 pp 1808ndash18272012
[44] H Song L Yu D Zhang and W-A Zhang ldquoFinite-time 119867infin
control for a class of discrete-time switched time-delay systemswith quantized feedbackrdquoCommunications in Nonlinear Scienceand Numerical Simulation vol 17 no 12 pp 4802ndash4814 2012
[45] Z Zuo H Li Y Liu and Y Wang ldquoOn finite-time stochasticstability and stabilization ofMarkovian jump systems subject topartial information on transition probabilitiesrdquoCircuits Systemsand Signal Processing 2012
[46] H Liu Y Shen and X Zhao ldquoDelay-dependent observer-based119867
infinfinite-time control for switched systems with time-varying
delayrdquoNonlinear Analysis Hybrid Systems vol 6 no 3 pp 885ndash898 2012
[47] Y Zhang C Liu and X Mu ldquoRobust finite-time119867infincontrol of
singular stochastic systems via static output feedbackrdquo AppliedMathematics and Computation vol 218 no 9 pp 5629ndash56402012
Journal of Applied Mathematics 13
[48] J Cheng H Zhu S Zhong Y Zhang and Y Zeng ldquoFinite-timeStabilization of 119867
infinfiltering for switched stochastic systemsrdquo
Circuits Systems and Signal Processing 2012[49] J Cheng H Zhu S Zhong and Y Zhang ldquoFinite-time bound-
ness of 119867infin
filtering for switching discrete-time systemsrdquo In-ternational Journal of Control Automation and Systems vol 10pp 1129ndash1135 2012
[50] Z Xiang C Qiao and M S Mahmoud ldquoFinite-time analysisand119867
infincontrol for switched stochastic systemsrdquo Journal of the
Franklin Institute vol 349 no 3 pp 915ndash927 2012
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Journal of Applied Mathematics 7
From Lemma 5 it yields that
minus int
119905
119905minusℎ
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
= minusint
119905
119905minus120591119898(119905)
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
minus int
119905minus120591119898(119905)
119905minusℎ
120578⊺
(119904) 119884119894119898120578 (119904) 119889119904
le minus
ℎ
120591119898(119905)
[int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904]
⊺
119884119894119898
ℎ
[int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904]
minus
ℎ
ℎ minus 120591119898(119905)
[int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904]
⊺
119884119894119898
ℎ
[int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904]
le minus
[
[
[
[
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904
]
]
]
]
⊺
[
[
[
119884119894119898
ℎ
119878119894119898
lowast
119884119894119898
ℎ
]
]
]
[
[
[
[
int
119905
119905minus120591119898(119905)
120578 (119904) 119889119904
int
119905minus120591119898(119905)
119905minusℎ
120578 (119904) 119889119904
]
]
]
]
(37)
From (25)ndash(37) we can eventually obtain
pound119881 (120578119905 119903
119905 119904
119905) minus 120575120596
⊺
(119905)119867120596 (119905) le 119890120575119905
120585⊺
(119905) Ξ119894119898120585 (119905) (38)
where
120585⊺
(119905) = [120578⊺
(119905) 120578⊺
(119905 minus 120591119898(119905)) 120578
⊺
(119905 minus ℎ) 120578⊺
(119905) 119880⊺
1 119880
⊺
2 120596
⊺
(119905)]
Ξ119894119898=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860⊺
119894119881
⊺
119894119898120591119898(119905) 119860
⊺
119894119872
⊺
119894119898(ℎ minus 120591
119898(119905)) 119860
⊺
119894119873
⊺
119894119898119875119894119898119863
119894
lowast Ξ22119894119898
119878119894119898minus
119884119894119898
ℎ
119860⊺
120591119894119881
⊺
119894119898120591119898(119905) 119860
⊺
120591119894119872
⊺
119894119898(ℎ minus 120591
119898(119905)) 119860
⊺
120591119894119873
⊺
1198941198980
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minus120591119898(119905)119872
⊺
119894119898minus (ℎ minus 120591
119898(119905))119873
⊺
119894119898119881119894119898119863
119894
lowast lowast lowast lowast minus120591119898(119905) 119883
1198941198980 120591
119898(119905)119872
119894119898119863
119894
lowast lowast lowast lowast lowast minus (ℎ minus 120591119898(119905))119883
119894119898(ℎ minus 120591
119898(119905))119873
119894119898119863
119894
lowast lowast lowast lowast lowast lowast minus120575119867
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
]
(39)
The LMIs (20) and (21) lead to 120591119898(119905) rarr ℎ and 120591
119898(119905) rarr
0 respectively It is easy to see that Ξ1119894119898
results from Ξ119894119898|
120591119898(119905) = ℎ and Ξ
119894119898|120591119898(119905) = 0 Thus we obtain
E pound119881 (120578119905 119903
119905 119904
119905) le 120578E [119881 (120578
119905 119903
119905 119904
119905)] + 120575120596
⊺
(119905)119867120596 (119905)
(40)
Multiplying the aforementioned inequality by 119890minus120578119905 we canget
E pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] le 119890
minus120578119905
120575120596⊺
(119905)119867120596 (119905) (41)
By integrating the aforementioned inequality between 0and 119905 it follows that
119890minus120578119905
E [119881 (120578119905 119903
119905 119904
119905)] minus E [119881 (120578
0 119903
0 119904
0)]
le 120575int
119905
0
119890minus120578119904
120596⊺
(119904)119867120596 (119904) 119889119904
(42)
Denote that 119894119898
= 119877minus(12)
119875119894119898119877minus(12) 119876
119904119894=
119877minus(12)
119876119904119894119877minus(12)
(119904 = 1 2) 119876 = 119877minus(12)
119876119877minus(12)
119883119894119898
= 119877minus(12)
119883119894119898119877minus(12) 119883 = 119877
minus(12)
119883119877minus(12)
119894119898
= 119877minus(12)
119884119894119898119877minus(12) and = 119877
minus(12)
119884119877minus(12) it yields
that
E [119881 (1205780 119903
0 119904
0)]
le max119894isinN119898isinM
120582max (119894119898) 120578⊺
(0) 119877120578 (0)
+ (max119894isinN
120582max (1198761119894) +max
119894isinN120582max (1198762119894
)) 119890120575ℎ
times int
0
minusℎ
119890120575119904
120578⊺
(119904) 119877120578 (119904) 119889119904
+ 119890120575ℎ
120582max (119876)int0
minusℎ
int
0
120579
119890120575119904
120578⊺
(119904) 119877120578 (119904) 119889119904
+ 119890120575ℎ max
119894isinN119898isinM120582max (119883119894119898
)
times int
0
minusℎ
int
0
120579
119890minus120575120579
120578⊺
(119904) 119877120578 (119904) 119889119904 119889120579
+ 119890120575ℎ
120582max (119883)
times int
0
minusℎ
int
0
120579
int
0
120592
119890minus120575120579
120578⊺
(119904) 119877120578 (119904) 119889119904 119889120579 119889120592
+ 119890120575ℎ max
119894isinN119898isinM120582max (119894119898)
8 Journal of Applied Mathematics
times int
0
minusℎ
int
0
120579
119890minus120575120579
120578⊺
(119904) 119877 120578 (119904) 119889119904 119889120579
+ 119890120575ℎ
120582max ()
times int
0
minusℎ
int
0
120579
int
0
120592
119890minus120575120579
120578⊺
(119904) 119877 120578 (119904) 119889119904 119889120579 119889120592
le max119894isinN119898isinM
120582max (119894119898)
+ ℎ119890120575ℎ
(max119894isinN
120582max (1198761119894+
119894isinNmax120582max (1198762119894
)))
+ ℎ2
119890120575ℎ
( max119894isinN119898isinM
120582max (119883119894119898) + 120582max (119876)
+ max119894isinN119898isinM
120582max (119894119898))
+
1
2
ℎ3
119890120575ℎ
(120582max (119883) + 120582max ())
times supminusℎle119904le0
120578⊺
(119904) 119877120578 (119904) 120578⊺
(119904) 119877 120578 (119904)
= 1198881Λ
(43)
For given 120578 gt 0 and 0 le 119905 le 119879 we have
E [119881 (120578119905 119903
119905 119904
119905)] le E [119890
120578119905
119881 (1205780 119903
0 119904
0)]
+ 119890120578119905
120575int
119905
0
119890minus120578119904
120596⊺
(119904)119867120596 (119904) 119889119904
le 119890120578119879
1198881Λ + 119889120575119890
120578119879
120582max (119867)int119879
0
119890minus120578119904
119889119904
le 119890120578119879
1198881Λ + 119889120575120582
10
1
120578
(1 minus 119890minus120578119879
)
(44)
On the other hand it follows from (25) that
E [119881 (120578119905 119903
119905 119904
119905)] ge E [120578
⊺
(119905) 119890120582119905
119875119894119898120578 (119905)]
ge max119894isinN
120582min (119875119894119898)E [120578⊺
(119905) 119877120578 (119905)]
= 1205821E [120578
⊺
(119905) 119877120578 (119905)]
(45)
It can be derived from (43)ndash(45) that
E [120578⊺
(119905) 119877120578 (119905)] le
119890120578119879
1205821
1198881Λ + 119889120575120582
10
1
120578
(1 minus 119890minus120578119879
) (46)
From (22) and (46) we have
E [120578⊺
(119905) 119877120578 (119905)] lt 1198882 (47)
Then the system is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879)
Remark 8 In this paper 120591119898(119905) and 120591
119898(119905) may have different
upper bounds in various delay intervals satisfying (7) respec-tively However in previous work such as [20 21] 120591
119898(119905) and
120591119898(119905) are enlarged to 120591
119898(119905) le ℎ = maxℎ
119894 119898 isin M and
120591119898(119905) le 120583 = max120583
119898 119898 isin M respectively which may lead
to conservativeness inevitably However the previouse casecan be taken fully into account by employing the Lyapunov-Krasovskii functional (25)
Remark 9 When dealing with termminusint
119905
119905minusℎ
120578⊺
(119904)119884119894119898120578(119904)119889119904 the
convex combination is not employed Lemma 5 is used in thispaper and then the free-weighting matrices-dependent nulladd items are necessary to be introduced in our proof whichlead to the decrease of the number of LMIs and LMIs scalardecision variables
Remark 10 The feature of this paper is the way to dealwith the integral term Many researchers have enlarged thederivative of the Lyapunov functional in order to deal with theintegral term in mathematical operations In this paper wepropose a novel delay-dependent sufficient criterion whichensures that the Markovian jump system with different modesystems is finite-time stable
Remark 11 It should be pointed out that the novelty of theLyapunov functional (25) lies in distinct Lyapunov matrices(119875
119894119898 119883
119894119898 119884
119894119898) which is chosen for different system modes
119894 (119894 = 1 2 119873) and119898 (119898 = 1 2 119872)
Theorem 12 System (9) is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879) if there exist matrices 119875
119894gt 0 119876
119897119894gt 0 (119897 = 1 2)
119876 gt 0 119883119894119898gt 0 119883 gt 0 119884
119894119898gt 0 119884 gt 0 119878
119894119898 scalars 119888
1lt 119888
2
119879 gt 0 120582119904gt 0 (119904 = 1 2 9) 120578 gt 0 120574 gt 0 and Λ gt 0 such
that for all 119894 119895 isinN and119898 119899 isinM and (16)ndash(19) the followinginequalities hold
Σ1119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119872⊺
119894119898119875119894119898119863119894
119862
⊺
119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119872⊺
1198941198980 119864
⊺
119894
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119872⊺
119894119898119881119894119898119863119894
0
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119872119894119898119863119894
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(48)
Σ2119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119873⊺
119894119898119875119894119898119863119894
119862
⊺
119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119873⊺
1198941198980 119864
⊺
119894
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119873⊺
119894119898119881119894119898119863119894
0
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119873119894119898119863119894
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(49)
1198881Λ + 119889120575120574
21
120578
(1 minus 119890minus120578119879
) lt 1205821119890minus120578119879
1198882 (50)
Journal of Applied Mathematics 9
Proof We now consider the 119867infin
performance of system (9)Select the same Lyapunov-Krasovskii functional as Theo-rem 7 it yields that
pound119881 (120578119905 119903
119905 119904
119905) + 119890
⊺
(119905) 119890 (119905) minus 1205742
120596⊺
(119905) 120596 (119905) le 120585⊺
(119905) Σ119897119894119898120585 (119905)
(119897 = 1 2)
(51)
It follows from (49)-(50) that
E pound119881 (120578119905 119903
119905 119904
119905)
le E [120578119881 (120578119905 119903
119905 119904
119905)] + 120574
2
120596⊺
(119905) 120596 (119905) minus E [119890⊺
(119905) 119890 (119905)]
(52)
Multiplying the aforementioned inequality by 119890minus120578119905 onehas
E pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] le 119890
minus120578119905
[1205742
120596⊺
(119905) 120596 (119905) minus 119890⊺
(119905) 119890 (119905)]
(53)
In zero initial condition and E[119881(120578119905 119903
119905 119904
119905)] gt 0 by inte-
grating the aforementioned inequality between 0 and 119879 wecan get
int
119879
0
119890minus120578119905
[1205742
120596⊺
(119905) 120596 (119905) minus 119890⊺
(119905) 119890 (119905)] 119889119905
le Eint119879
0
pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] 119889119905 le 119881 (120578
0 119903
0 119904
0) = 0
(54)
Using Dynkins formula it results that
E [int119879
0
119890minus120578120592
119890⊺
(120592) 119890 (120592) 119889120592] le 1205742
E [int119879
0
119890minus120578120592
120596⊺
(120592) 120596 (120592) 119889120592]
(55)
Then it yields that
E [int119879
0
119890⊺
(120592) 119890 (120592) 119889120592] le 1205742
119890120578119879
E [int119879
0
120596⊺
(120592) 120596 (120592) 119889120592] (56)
Thus it is concluded by Definition 3 that system (9) is fi-nite-time bounded with an 119867
infinperformance 120574 This com-
pletes the proof
Remark 13 From the proof process of Theorems 7 and 12it is easy to see that neither bounding technique for cross-terms nor model transformation is involved In other wordsthe obtained result is expected to be less conservative
Remark 14 The Lyapunov asymptotic stability and finite-time stability of a class of system are independent conceptsA Lyapunov asymptotic stability system may not be finite-time stability Moreover finite-time stability system may alsonot be Lyapunov asymptotic stabilityThere exist some resultson Lyapunov stability while finite-time stability also needsour full investigation which was neglected by most previouswork
4 Finite-Time 119867infin
Filtering
Theorem 15 System (9) is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879) if there exist matrices 119875
119894119898gt 0119872
119894119898 119873
119894119898 119881
119894119898
119876119897119894gt 0 (119897 = 1 2) 119876 gt 0 119883
119894119898gt 0 119883 gt 0 119884
119894119898gt 0 119884 gt 0
119860119891119894119898
119861119891119894119898
119862119891119894119898
119872119860119894119898
119872119861119894119898
119873119860119894119898
119873119861119894119898
119881119860119894119898
119881119861119894119898
119878119894119898
scalars 1198881lt 119888
2 119879 gt 0 120590
119904gt 0 (119904 = 1 2 9) 120575 gt 0 120578 gt 0
and Λ gt 0 such that for all 119894 119895 isin N and 119898 119899 isin M thefollowing inequalities hold
119875119894119898= [
1198751119894119898
1198752119894119898
1198752119894119898
1198752119894119898
] gt 0 119872119894119898= [
1198721119894119898
1198722119894119898
1198722119894119898
1198722119894119898
] (57)
119873119894119898= [
1198731119894119898
1198732119894119898
1198732119894119898
1198732119894119898
] 119881119894119898= [
1198811119894119898
1198812119894119898
1198812119894119898
1198812119894119898
] (58)
Γ1119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Γ11119894119898
Γ12119894119898
minus119878119894119898
Γ14119894119898
Γ15119894119898
Γ16119894119898
Γ17119894119898
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
Γ24119894119898
Γ25119894119898
0 [119862⊺
119911120591119894
0]
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119872⊺
119894119898Γ46119894119898
0
lowast lowast lowast lowast minusℎ119883119894119898
Γ56119894119898
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(59)
Γ2119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Γ11119894119898
Γ12119894119898
minus119878119894119898
Γ14119894119898
Γ1015840
15119894119898Γ16119894119898
Γ17119894119898
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
Γ24119894119898
Γ1015840
251198941198980 [
119862⊺
119911120591119894
0]
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119873⊺
119894119898Γ1015840
461198941198980
lowast lowast lowast lowast minusℎ119883119894119898
Γ1015840
561198941198980
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(60)where
Γ11119894119898
= sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898+ 119890
120575ℎ
1198761119894+ 119890
120575ℎ
1198762119894+ ℎ119876 +
119890120575ℎ
minus 1
120575
119883119894119898+
119890120575ℎ
minus 120575ℎ119890120575ℎ
minus 1
1205752
119883
+
119884119894119898
ℎ
+ [
1205751198751119894119898
+ 1198751119894119898119860
119894+ 119860
⊺
1198941198751119894119898
+ 119861119891119894119898119862119910119894+ 119862
⊺
119910119894119861
⊺
119891119894119898120575119875
2119894119898+ 119860
119891119894119898+ 119860
⊺
119891119894119898
1205751198752119894119898
+ 1198752119894119898119860
119894+ 119860
⊺
1198941198752119894119898
+ 119861119891119894119898119862119910119894+ 119862
⊺
119910119894119861
⊺
119891119894119898120575119875
2119894119898+ 119860
119891119894119898+ 119860
⊺
119891119894119898
]
(61)
10 Journal of Applied Mathematics
Γ12119894119898
= minus
119884119894119898
ℎ
+ 119878119894119898+ [
1198751119894119898119860
120591119894+ 119861
119891119894119898119862119910120591119894
1198752119894119898119860
120591119894+ 119861
119891119894119898119862119910120591119894
]
Γ14119894119898
= [
119860⊺
119894119881
⊺
1119894119898+ 119862
⊺
119910119894119881
⊺
119861119894119898119881
⊺
119860119894119898
119860⊺
119894119881
⊺
2119894119898+ 119862
⊺
119910119894119881
⊺
119861119894119898119881
⊺
119860119894119898
]
Γ15119894119898
= [
ℎ119860⊺
119894119872
⊺
1119894119898+ ℎ119862
⊺
119910119894119872
⊺
119861119894119898ℎ119872
⊺
119860119894119898
ℎ119860⊺
119894119872
⊺
2119894119898+ ℎ119862
⊺
119910119894119872
⊺
119861119894119898ℎ119872
⊺
119860119894119898
]
Γ1015840
15119894119898= [
ℎ119860⊺
119894119873
⊺
1119894119898+ ℎ119862
⊺
119910119894119873
⊺
119861119894119898ℎ119873
⊺
119860119894119898
ℎ119860⊺
119894119873
⊺
2119894119898+ ℎ119862
⊺
119910119894119873
⊺
119861119894119898ℎ119873
⊺
119860119894119898
]
Γ16119894119898
= [
1198751119894119898119863
119894+ 119861
119891119894119898119863
119910119894
1198752119894119898119863
119894+ 119861
119891119894119898119863
119910119894
]
Γ17119894119898
= [
119862⊺
119911119894
minus119862119891119894119898
]
Γ24119894119898
= [
119860⊺
120591119894119881
⊺
1119894119898+ 119862
⊺
119910120591119894119881
⊺
119861119894119898
119860⊺
120591119894119881
⊺
2119894119898+ 119862
⊺
119910120591119894119881
⊺
119861119894119898
]
Γ25119894119898
= [
ℎ119860⊺
120591119894119872
⊺
1119894119898+ ℎ119862
⊺
119910120591119894119872
⊺
119861119894119898
ℎ119860⊺
120591119894119872
⊺
2119894119898+ ℎ119862
⊺
119910120591119894119872
⊺
119861119894119898
]
Γ1015840
25119894119898= [
ℎ119860⊺
120591119894119873
⊺
1119894119898+ ℎ119862
⊺
119910120591119894119873
⊺
119861119894119898
ℎ119860⊺
120591119894119873
⊺
2119894119898+ ℎ119862
⊺
119910120591119894119873
⊺
119861119894119898
]
Γ46119894119898
= [
1198811119894119898119863
119894+ 119881
119861119894119898119863
119910119894
1198812119894119898119863
119894+ 119881
119861119894119898119863
119910119894
]
Γ56119894119898
= [
ℎ1198721119894119898119863
119894+ ℎ119872
119861119894119898119863
119910119894
ℎ1198722119894119898119863
119894+ ℎ119872
119861119894119898119863
119910119894
]
Γ1015840
56119894119898= [
ℎ1198731119894119898119863
119894+ ℎ119873
119861119894119898119863
119910119894
ℎ1198732119894119898119863
119894+ ℎ119873
119861119894119898119863
119910119894
]
(62)
Then a desired filter can be chosen with parameters as
119860119891119894119898
= 119875minus1
2119894119898119860
119891119894119898 119861
119891119894119898= 119875
minus1
2119894119898119861119891119894119898
119862119891119894119898
= 119862119891119894119898
(63)
Proof We denote that
119875119894119898= [
1198751119894119898
1198752119894119898
1198752119894119898
1198752119894119898
] (64)
The term 119875119894119898119860
119894can be rewritten as
119875119894119898119860
119894= [
1198751119894119898119860
119894+ 119875
21198941198981198611198911198941198981198621199101198941198752119894119898119860
119891119894119898
1198752119894119898119860
119894+ 119875
21198941198981198611198911198941198981198621199101198941198752119894119898119860
119891119894119898
] (65)
Similarly we have
119875119894119898119861119894= [
1198751119894119898119860
120591119894+ 119875
2119894119898119861119891119894119898119862119910120591119894
1198752119894119898119860
120591119894+ 119875
2119894119898119861119891119894119898119862119910120591119894
]
119875119894119898119863
119894= [
1198751119894119898119863
119894+ 119875
2119894119898119861119891119894119898119863
119910119894
1198752119894119898119863
119894+ 119875
2119894119898119861119891119894119898119863
119910119894
]
(66)
Define 119860119891119894119898
= 1198752119894119898119860
119891119894119898 119861
119891119894119898= 119875
2119894119898119861119891119894119898
119862119891119894119898
= 119862119891119894119898
119872
119860119894119898= 119872
2119894119898119860
119891119894119898119872
119861119894119898= 119872
2119894119898119861119891119894119898
119873119860119894119898
= 1198732119894119898119860
119891119894119898
119873119861119894119898
= 1198732119894119898119861119891119894119898
119881119860119894119898
= 1198812119894119898119860
119891119894119898 and 119881
119861119894119898= 119881
2119894119898119861119891119894119898
Therefore if (59) and (60) hold system (9) is finite-timebounded with a prescribed 119867
infinperformance index 120574 The
proof is completed
Remark 16 In many actual applications the minimum valueof 1205742min is of interest In Theorem 12 with a fixed 120582 120574min canbe obtained through the following optimization procedure
min 1205742
st (48)ndash(50) (67)
In Theorem 15 as for finite-time stability and bounded-ness once the state bound 119888
2is not ascertained theminimum
value 1198882min is of interest With a fixed 120582 define 120582
1= 1 then
the following optimization problem can be formulated to getminimum value 119888
2min
min 1205891205742
+ (1 minus 120589) 1198882
st (59)-(60) and (50)
(68)
where 120589 is weighted factor and 120589 isin [0 1]
5 Illustrative Example
Example 17 Consider that the Markovian jump system andthe delay mode switching are governed by a Markov processwith the following transition rates
Λ = [
minus07 07
09 minus09] Π = [
minus1 1
12 minus12] (69)
as well as with the following parameters
1198601= [
minus35 086
minus064 minus325] 119860
2= [
minus25 034
14 minus002]
1198601205911= [
minus08 minus13
minus07 minus22] 119860
1205912= [
minus28 05
minus08 minus14]
1198631= 119863
2= [
01
02] 119862
1199101= [04 07]
1198621199102= [05 08] 119862
1199101205911= [01 02]
1198621199101205912
= [01 01]
1198631199101= 119863
1199102= 01 119862
1199111= [01 005]
1198621199112= [02 009] 119862
1199111205911= [03 06]
1198621199111205912
= [04 08] 1198631199111= 119863
1199112= 005
(70)
Journal of Applied Mathematics 11
Then we choose 119877 = 119868 119879 = 2 1198881= 1 and 119889 = 001
throughTheorem 15 it yields that themode-dependent filtersare as follows
11986011989111
= [
minus61254 15565
minus02347 minus03763]
11986011989112
= [
minus62682 14338
minus03552 minus05327]
11986011989121
= [
minus89214 24223
minus15476 minus06234]
11986011989122
= [
minus84638 28237
minus13545 minus04322]
11986111989111
= [
51465 minus28516
minus91216 95171]
11986111989112
= [
54203 minus23121
minus93156 96332]
11986111989121
= [
46193 minus11984
minus162397 163006]
11986111989122
= [
46512 minus10311
minus161132 162312]
11986211989111
= [00294 minus00397]
11986211989112
= [00271 minus003368]
11986211989121
= [01163 minus01432]
11986211989122
= [01342 minus03672]
(71)
This paper deals with the finite-time filter design problemfor a class of Markovian jump systems particularly two dif-ferentMarkov processes are considered formodeling the ran-domness of system matrix and the state delay Then throughthe numerical example we can see that results in this paperare feasible which further verified the correctness of our the-ory Therefore the paper shorten this gap
6 Conclusions
In this paper we have examined the problems of finite-time119867
infinfiltering for a class of Markovian jump systems with dif-
ferent system modes Based on a novel approach a sufficientcondition is derived such that the closed-loop Markovianjump system is finite-time bounded and satisfies a prescribedlevel of119867
infindisturbance attenuation in a finite time interval
Finally a numerical example is also given to illustrate theeffectiveness of the proposed design approach It should benoted that one of future research topics would be to investi-gate the problems of fault detection and fault tolerant controlfor time-varying Markovian jump systems with incompleteinformation over a finite-time horizon
Acknowledgments
The authors would like to thank the associate editor and theanonymous reviewers for their detailed comments and sug-gestions This work was supported by the Fund of SichuanProvincial Key Laboratory of Signal and Information Pro-cessing Xihua University (SZJJ2009-002 and SGXZD0101-10-1) and National Basic Research Program of China(2010CB732501)
References
[1] N N Krasovskiı and E A Lidskiı ldquoAnalytical design ofcontrollers in systems with random attributesmdashI Statementof the problem method of solvingrdquo Automation and RemoteControl vol 22 pp 1021ndash1025 1961
[2] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[3] L Wu X Su and P Shi ldquoSliding mode control with bounded1198712gain performance of Markovian jump singular time-delay
systemsrdquo Automatica vol 48 no 8 pp 1929ndash1933 2012[4] H Gao Z Fei J Lam and B Du ldquoFurther results on expo-
nential estimates of Markovian jump systems with mode-de-pendent time-varying delaysrdquo IEEE Transactions on AutomaticControl vol 56 no 1 pp 223ndash229 2011
[5] Z Wu H Su and J Chu ldquo119867infinfiltering for singular Markovian
jump systems with time delayrdquo International Journal of Robustand Nonlinear Control vol 20 no 8 pp 939ndash957 2010
[6] P Tino M Cernansky and L Benuskova ldquoMarkovian architec-tural bias of recurrent neural networksrdquo IEEE Transactions onNeural Networks vol 15 no 1 pp 6ndash15 2004
[7] H Dong Z Wang D W C Ho and H Gao ldquoRobust 119867infin
filtering for Markovian jump systems with randomly occurringnonlinearities and sensor saturation the finite-horizon caserdquoIEEE Transactions on Signal Processing vol 59 no 7 pp 3048ndash3057 2011
[8] Q Ma S Xu and Y Zou ldquoStability and synchronization forMarkovian jump neural networks with partly unknown tran-sition probabilitiesrdquo Neurocomputing vol 74 pp 3404ndash33412011
[9] H Zhao S Xu and Y Zou ldquoRobust119867infinfiltering for uncertain
Markovian jump systems with mode-dependent distributeddelaysrdquo International Journal of Adaptive Control and SignalProcessing vol 24 no 1 pp 83ndash94 2010
[10] Y Zhang S Xu and B Zhang ldquoDiscrete-time fuzzy Markovianjump systems with time-varying delaysrdquo IEEE Transactions onFuzzy Systems vol 17 no 2 pp 411ndash420 2009
[11] Z Wu H Su and J Chu ldquoState estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 no 10ndash12 pp 2247ndash2254 2010
[12] Q Zhu and J Cao ldquoRobust exponential stability of markovianjump impulsive stochastic Cohen-Grossberg neural networkswithmixed time delaysrdquo IEEE Transactions onNeural Networksvol 21 no 8 pp 1314ndash1325 2010
[13] H Huang G Feng and X Chen ldquoStability and stabilizationof Markovian jump systems with time delay via new Lyapunovfunctionalsrdquo IEEE Transactions on Circuits and Systems I vol59 no 10 pp 2413ndash2421 2012
12 Journal of Applied Mathematics
[14] Q Zhu and J Cao ldquoStability of Markovian jump neural net-works with impulse control and time varying delaysrdquoNonlinearAnalysis Real World Applications vol 13 no 5 pp 2259ndash22702012
[15] M Sun J Lam S Xu andY Zou ldquoRobust exponential stabiliza-tion for Markovian jump systems with mode-dependent inputdelayrdquo Automatica vol 43 no 10 pp 1799ndash1807 2007
[16] M Mohammadian A Abolmasoumi and H Momeni ldquo119867infin
mode-independent filter design for Markovian jump geneticregulatory networks with time-varying delaysrdquo Neurocomput-ing vol 87 pp 10ndash18 2012
[17] S Boyd L El Ghaoui E Feron and V Balakrishnan LinearMatrix Inequalities in SystemandControlTheory vol 15 of SIAMStudies in Applied Mathematics SIAM Philadelphia Pa USA1994
[18] S Mou H GaoW Qiang and K Chen ldquoNew delay-dependentexponential stability for neural networks with time delayrdquo IEEETransactions on Systems Man and Cybernetics Part B vol 38no 2 pp 571ndash576 2008
[19] O M Kwon J Park S Lee and E Cha ldquoAnalysis on delay-dependent stability for neural networks with time-varyingdelaysrdquo Neurocomputing vol 103 pp 114ndash120 2013
[20] L Hu H Gao andW X Zheng ldquoNovel stability of cellular neu-ral networkswith interval time-varying delayrdquoNeuralNetworksvol 21 no 10 pp 1458ndash1463 2008
[21] H Zhang Z Liu G B Huang and ZWang ldquoNovel weighting-delay-based stability criteria for recurrent neural networks withtime-varying delayrdquo IEEE Transactions on Neural Networks vol21 no 1 pp 91ndash106 2010
[22] Z Zuo C Yang and YWang ldquoA newmethod for stability anal-ysis of recurrent neural networks with interval time-varyingdelayrdquo IEEE Transactions on Neural Networks vol 21 no 2 pp339ndash344 2010
[23] P Park JWKo andC Jeong ldquoReciprocally convex approach tostability of systems with time-varying delaysrdquo Automatica vol47 no 1 pp 235ndash238 2011
[24] D Zhang L Yu and W Zhang ldquoDelay-dependent fault detec-tion for switched linear systems with time-varying delays-theaverage dwell time approachrdquo Signal Processing vol 91 pp 832ndash840 2011
[25] Z Fei H Gao and P Shi ldquoNew results on stabilization ofMark-ovian jump systems with time delayrdquoAutomatica vol 45 no 10pp 2300ndash2306 2009
[26] H R Karimi ldquoRobust delay-dependent 119867infin
control of uncer-tain time-delay systems with mixed neutral discrete anddistributed time-delays and Markovian switching parametersrdquoIEEE Transactions on Circuits and Systems I vol 58 no 8 pp1910ndash1923 2011
[27] M Luo G Liu and S Zhong ldquoRobust fault detection ofMarkovian jump systems with different systemmodesrdquoAppliedMathematical Modelling vol 37 no 7 pp 5001ndash5012 2013
[28] B Zhang andY Li ldquoExponential1198712minus119871
infinfiltering for distributed
delay systems with Markovian jumping parametersrdquo SignalProcessing vol 93 no 1 pp 206ndash216 2013
[29] A Abolmasoumi and H Momeni ldquoRobust observer-based119867infin
control of a Markovian jump system with different delay andsystem modesrdquo International Journal of Control Automationand Systems vol 9 no 4 pp 768ndash776 2011
[30] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[31] S Ma and E K Boukas ldquoRobust 119867infin
filtering for uncertaindiscrete Markov jump singular systems with mode-dependenttime delayrdquo IET Control Theory amp Applications vol 3 no 3 pp351ndash361 2009
[32] Y ChenW Bi andW Li ldquoStability analysis for neural networkswith time-varying delay a more general delay decompositionapproachrdquo Neurocomputing vol 73 no 4ndash6 pp 853ndash857 2010
[33] Z Shu J Lam and S Xu ldquoRobust stabilization of Markoviandelay systems with delay-dependent exponential estimatesrdquoAutomatica vol 42 no 11 pp 2001ndash2008 2006
[34] X Huang W Lin and B Yang ldquoGlobal finite-time stabilizationof a class of uncertain nonlinear systemsrdquo Automatica vol 41no 5 pp 881ndash888 2005
[35] C Qian and J Li ldquoGlobal finite-time stabilization by outputfeedback for planar systems without observable linearizationrdquoIEEE Transactions on Automatic Control vol 50 no 6 pp 885ndash890 2005
[36] F Amato R Ambrosino C Cosentino and G De TommasildquoInput-output finite time stabilization of linear systemsrdquo Auto-matica vol 46 no 9 pp 1558ndash1562 2010
[37] Y Hong ldquoFinite-time stabilization and stabilizability of a classof controllable systemsrdquo Systems amp Control Letters vol 46 no4 pp 231ndash236 2002
[38] F Amato and M Ariola ldquoFinite-time control of discrete-timelinear systemsrdquo IEEETransactions onAutomatic Control vol 50no 5 pp 724ndash729 2005
[39] Z Zuo Y Liu Y Wang and H Li ldquoFinite-time stochasticstability and stabilisation of linear discrete-time Markovianjump systems with partly unknown transition probabilitiesrdquoIET Control Theory amp Applications vol 6 no 10 pp 1522ndash15262012
[40] X Luan F Liu and P Shi ldquoFinite-time filtering for non-linearstochastic systems with partially known transition jump ratesrdquoIET Control Theory amp Applications vol 4 no 5 pp 735ndash7452010
[41] F Amato M Ariola and C Cosentino ldquoFinite-time control ofdiscrete-time linear systems analysis and design conditionsrdquoAutomatica vol 46 no 5 pp 919ndash924 2010
[42] X Lin H Du and S Li ldquoFinite-time boundedness and 1198712-
gain analysis for switched delay systems with norm-boundeddisturbancerdquo Applied Mathematics and Computation vol 217no 12 pp 5982ndash5993 2011
[43] J Lin S Fei and Z Gao ldquoStabilization of discrete-time switchedsingular time-delay systems under asynchronous switchingrdquoJournal of the Franklin Institute vol 349 no 5 pp 1808ndash18272012
[44] H Song L Yu D Zhang and W-A Zhang ldquoFinite-time 119867infin
control for a class of discrete-time switched time-delay systemswith quantized feedbackrdquoCommunications in Nonlinear Scienceand Numerical Simulation vol 17 no 12 pp 4802ndash4814 2012
[45] Z Zuo H Li Y Liu and Y Wang ldquoOn finite-time stochasticstability and stabilization ofMarkovian jump systems subject topartial information on transition probabilitiesrdquoCircuits Systemsand Signal Processing 2012
[46] H Liu Y Shen and X Zhao ldquoDelay-dependent observer-based119867
infinfinite-time control for switched systems with time-varying
delayrdquoNonlinear Analysis Hybrid Systems vol 6 no 3 pp 885ndash898 2012
[47] Y Zhang C Liu and X Mu ldquoRobust finite-time119867infincontrol of
singular stochastic systems via static output feedbackrdquo AppliedMathematics and Computation vol 218 no 9 pp 5629ndash56402012
Journal of Applied Mathematics 13
[48] J Cheng H Zhu S Zhong Y Zhang and Y Zeng ldquoFinite-timeStabilization of 119867
infinfiltering for switched stochastic systemsrdquo
Circuits Systems and Signal Processing 2012[49] J Cheng H Zhu S Zhong and Y Zhang ldquoFinite-time bound-
ness of 119867infin
filtering for switching discrete-time systemsrdquo In-ternational Journal of Control Automation and Systems vol 10pp 1129ndash1135 2012
[50] Z Xiang C Qiao and M S Mahmoud ldquoFinite-time analysisand119867
infincontrol for switched stochastic systemsrdquo Journal of the
Franklin Institute vol 349 no 3 pp 915ndash927 2012
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
8 Journal of Applied Mathematics
times int
0
minusℎ
int
0
120579
119890minus120575120579
120578⊺
(119904) 119877 120578 (119904) 119889119904 119889120579
+ 119890120575ℎ
120582max ()
times int
0
minusℎ
int
0
120579
int
0
120592
119890minus120575120579
120578⊺
(119904) 119877 120578 (119904) 119889119904 119889120579 119889120592
le max119894isinN119898isinM
120582max (119894119898)
+ ℎ119890120575ℎ
(max119894isinN
120582max (1198761119894+
119894isinNmax120582max (1198762119894
)))
+ ℎ2
119890120575ℎ
( max119894isinN119898isinM
120582max (119883119894119898) + 120582max (119876)
+ max119894isinN119898isinM
120582max (119894119898))
+
1
2
ℎ3
119890120575ℎ
(120582max (119883) + 120582max ())
times supminusℎle119904le0
120578⊺
(119904) 119877120578 (119904) 120578⊺
(119904) 119877 120578 (119904)
= 1198881Λ
(43)
For given 120578 gt 0 and 0 le 119905 le 119879 we have
E [119881 (120578119905 119903
119905 119904
119905)] le E [119890
120578119905
119881 (1205780 119903
0 119904
0)]
+ 119890120578119905
120575int
119905
0
119890minus120578119904
120596⊺
(119904)119867120596 (119904) 119889119904
le 119890120578119879
1198881Λ + 119889120575119890
120578119879
120582max (119867)int119879
0
119890minus120578119904
119889119904
le 119890120578119879
1198881Λ + 119889120575120582
10
1
120578
(1 minus 119890minus120578119879
)
(44)
On the other hand it follows from (25) that
E [119881 (120578119905 119903
119905 119904
119905)] ge E [120578
⊺
(119905) 119890120582119905
119875119894119898120578 (119905)]
ge max119894isinN
120582min (119875119894119898)E [120578⊺
(119905) 119877120578 (119905)]
= 1205821E [120578
⊺
(119905) 119877120578 (119905)]
(45)
It can be derived from (43)ndash(45) that
E [120578⊺
(119905) 119877120578 (119905)] le
119890120578119879
1205821
1198881Λ + 119889120575120582
10
1
120578
(1 minus 119890minus120578119879
) (46)
From (22) and (46) we have
E [120578⊺
(119905) 119877120578 (119905)] lt 1198882 (47)
Then the system is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879)
Remark 8 In this paper 120591119898(119905) and 120591
119898(119905) may have different
upper bounds in various delay intervals satisfying (7) respec-tively However in previous work such as [20 21] 120591
119898(119905) and
120591119898(119905) are enlarged to 120591
119898(119905) le ℎ = maxℎ
119894 119898 isin M and
120591119898(119905) le 120583 = max120583
119898 119898 isin M respectively which may lead
to conservativeness inevitably However the previouse casecan be taken fully into account by employing the Lyapunov-Krasovskii functional (25)
Remark 9 When dealing with termminusint
119905
119905minusℎ
120578⊺
(119904)119884119894119898120578(119904)119889119904 the
convex combination is not employed Lemma 5 is used in thispaper and then the free-weighting matrices-dependent nulladd items are necessary to be introduced in our proof whichlead to the decrease of the number of LMIs and LMIs scalardecision variables
Remark 10 The feature of this paper is the way to dealwith the integral term Many researchers have enlarged thederivative of the Lyapunov functional in order to deal with theintegral term in mathematical operations In this paper wepropose a novel delay-dependent sufficient criterion whichensures that the Markovian jump system with different modesystems is finite-time stable
Remark 11 It should be pointed out that the novelty of theLyapunov functional (25) lies in distinct Lyapunov matrices(119875
119894119898 119883
119894119898 119884
119894119898) which is chosen for different system modes
119894 (119894 = 1 2 119873) and119898 (119898 = 1 2 119872)
Theorem 12 System (9) is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879) if there exist matrices 119875
119894gt 0 119876
119897119894gt 0 (119897 = 1 2)
119876 gt 0 119883119894119898gt 0 119883 gt 0 119884
119894119898gt 0 119884 gt 0 119878
119894119898 scalars 119888
1lt 119888
2
119879 gt 0 120582119904gt 0 (119904 = 1 2 9) 120578 gt 0 120574 gt 0 and Λ gt 0 such
that for all 119894 119895 isinN and119898 119899 isinM and (16)ndash(19) the followinginequalities hold
Σ1119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119872⊺
119894119898119875119894119898119863119894
119862
⊺
119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119872⊺
1198941198980 119864
⊺
119894
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119872⊺
119894119898119881119894119898119863119894
0
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119872119894119898119863119894
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(48)
Σ2119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
Ξ11119894119898
Ξ12119894119898
minus119878119894119898
119860
⊺
119894119881⊺
119894119898ℎ119860
⊺
119894119873⊺
119894119898119875119894119898119863119894
119862
⊺
119894
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
119861
⊺
119894119881⊺
119894119898ℎ119861
⊺
119894119873⊺
1198941198980 119864
⊺
119894
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119873⊺
119894119898119881119894119898119863119894
0
lowast lowast lowast lowast minusℎ119883119894119898
ℎ119873119894119898119863119894
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(49)
1198881Λ + 119889120575120574
21
120578
(1 minus 119890minus120578119879
) lt 1205821119890minus120578119879
1198882 (50)
Journal of Applied Mathematics 9
Proof We now consider the 119867infin
performance of system (9)Select the same Lyapunov-Krasovskii functional as Theo-rem 7 it yields that
pound119881 (120578119905 119903
119905 119904
119905) + 119890
⊺
(119905) 119890 (119905) minus 1205742
120596⊺
(119905) 120596 (119905) le 120585⊺
(119905) Σ119897119894119898120585 (119905)
(119897 = 1 2)
(51)
It follows from (49)-(50) that
E pound119881 (120578119905 119903
119905 119904
119905)
le E [120578119881 (120578119905 119903
119905 119904
119905)] + 120574
2
120596⊺
(119905) 120596 (119905) minus E [119890⊺
(119905) 119890 (119905)]
(52)
Multiplying the aforementioned inequality by 119890minus120578119905 onehas
E pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] le 119890
minus120578119905
[1205742
120596⊺
(119905) 120596 (119905) minus 119890⊺
(119905) 119890 (119905)]
(53)
In zero initial condition and E[119881(120578119905 119903
119905 119904
119905)] gt 0 by inte-
grating the aforementioned inequality between 0 and 119879 wecan get
int
119879
0
119890minus120578119905
[1205742
120596⊺
(119905) 120596 (119905) minus 119890⊺
(119905) 119890 (119905)] 119889119905
le Eint119879
0
pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] 119889119905 le 119881 (120578
0 119903
0 119904
0) = 0
(54)
Using Dynkins formula it results that
E [int119879
0
119890minus120578120592
119890⊺
(120592) 119890 (120592) 119889120592] le 1205742
E [int119879
0
119890minus120578120592
120596⊺
(120592) 120596 (120592) 119889120592]
(55)
Then it yields that
E [int119879
0
119890⊺
(120592) 119890 (120592) 119889120592] le 1205742
119890120578119879
E [int119879
0
120596⊺
(120592) 120596 (120592) 119889120592] (56)
Thus it is concluded by Definition 3 that system (9) is fi-nite-time bounded with an 119867
infinperformance 120574 This com-
pletes the proof
Remark 13 From the proof process of Theorems 7 and 12it is easy to see that neither bounding technique for cross-terms nor model transformation is involved In other wordsthe obtained result is expected to be less conservative
Remark 14 The Lyapunov asymptotic stability and finite-time stability of a class of system are independent conceptsA Lyapunov asymptotic stability system may not be finite-time stability Moreover finite-time stability system may alsonot be Lyapunov asymptotic stabilityThere exist some resultson Lyapunov stability while finite-time stability also needsour full investigation which was neglected by most previouswork
4 Finite-Time 119867infin
Filtering
Theorem 15 System (9) is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879) if there exist matrices 119875
119894119898gt 0119872
119894119898 119873
119894119898 119881
119894119898
119876119897119894gt 0 (119897 = 1 2) 119876 gt 0 119883
119894119898gt 0 119883 gt 0 119884
119894119898gt 0 119884 gt 0
119860119891119894119898
119861119891119894119898
119862119891119894119898
119872119860119894119898
119872119861119894119898
119873119860119894119898
119873119861119894119898
119881119860119894119898
119881119861119894119898
119878119894119898
scalars 1198881lt 119888
2 119879 gt 0 120590
119904gt 0 (119904 = 1 2 9) 120575 gt 0 120578 gt 0
and Λ gt 0 such that for all 119894 119895 isin N and 119898 119899 isin M thefollowing inequalities hold
119875119894119898= [
1198751119894119898
1198752119894119898
1198752119894119898
1198752119894119898
] gt 0 119872119894119898= [
1198721119894119898
1198722119894119898
1198722119894119898
1198722119894119898
] (57)
119873119894119898= [
1198731119894119898
1198732119894119898
1198732119894119898
1198732119894119898
] 119881119894119898= [
1198811119894119898
1198812119894119898
1198812119894119898
1198812119894119898
] (58)
Γ1119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Γ11119894119898
Γ12119894119898
minus119878119894119898
Γ14119894119898
Γ15119894119898
Γ16119894119898
Γ17119894119898
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
Γ24119894119898
Γ25119894119898
0 [119862⊺
119911120591119894
0]
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119872⊺
119894119898Γ46119894119898
0
lowast lowast lowast lowast minusℎ119883119894119898
Γ56119894119898
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(59)
Γ2119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Γ11119894119898
Γ12119894119898
minus119878119894119898
Γ14119894119898
Γ1015840
15119894119898Γ16119894119898
Γ17119894119898
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
Γ24119894119898
Γ1015840
251198941198980 [
119862⊺
119911120591119894
0]
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119873⊺
119894119898Γ1015840
461198941198980
lowast lowast lowast lowast minusℎ119883119894119898
Γ1015840
561198941198980
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(60)where
Γ11119894119898
= sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898+ 119890
120575ℎ
1198761119894+ 119890
120575ℎ
1198762119894+ ℎ119876 +
119890120575ℎ
minus 1
120575
119883119894119898+
119890120575ℎ
minus 120575ℎ119890120575ℎ
minus 1
1205752
119883
+
119884119894119898
ℎ
+ [
1205751198751119894119898
+ 1198751119894119898119860
119894+ 119860
⊺
1198941198751119894119898
+ 119861119891119894119898119862119910119894+ 119862
⊺
119910119894119861
⊺
119891119894119898120575119875
2119894119898+ 119860
119891119894119898+ 119860
⊺
119891119894119898
1205751198752119894119898
+ 1198752119894119898119860
119894+ 119860
⊺
1198941198752119894119898
+ 119861119891119894119898119862119910119894+ 119862
⊺
119910119894119861
⊺
119891119894119898120575119875
2119894119898+ 119860
119891119894119898+ 119860
⊺
119891119894119898
]
(61)
10 Journal of Applied Mathematics
Γ12119894119898
= minus
119884119894119898
ℎ
+ 119878119894119898+ [
1198751119894119898119860
120591119894+ 119861
119891119894119898119862119910120591119894
1198752119894119898119860
120591119894+ 119861
119891119894119898119862119910120591119894
]
Γ14119894119898
= [
119860⊺
119894119881
⊺
1119894119898+ 119862
⊺
119910119894119881
⊺
119861119894119898119881
⊺
119860119894119898
119860⊺
119894119881
⊺
2119894119898+ 119862
⊺
119910119894119881
⊺
119861119894119898119881
⊺
119860119894119898
]
Γ15119894119898
= [
ℎ119860⊺
119894119872
⊺
1119894119898+ ℎ119862
⊺
119910119894119872
⊺
119861119894119898ℎ119872
⊺
119860119894119898
ℎ119860⊺
119894119872
⊺
2119894119898+ ℎ119862
⊺
119910119894119872
⊺
119861119894119898ℎ119872
⊺
119860119894119898
]
Γ1015840
15119894119898= [
ℎ119860⊺
119894119873
⊺
1119894119898+ ℎ119862
⊺
119910119894119873
⊺
119861119894119898ℎ119873
⊺
119860119894119898
ℎ119860⊺
119894119873
⊺
2119894119898+ ℎ119862
⊺
119910119894119873
⊺
119861119894119898ℎ119873
⊺
119860119894119898
]
Γ16119894119898
= [
1198751119894119898119863
119894+ 119861
119891119894119898119863
119910119894
1198752119894119898119863
119894+ 119861
119891119894119898119863
119910119894
]
Γ17119894119898
= [
119862⊺
119911119894
minus119862119891119894119898
]
Γ24119894119898
= [
119860⊺
120591119894119881
⊺
1119894119898+ 119862
⊺
119910120591119894119881
⊺
119861119894119898
119860⊺
120591119894119881
⊺
2119894119898+ 119862
⊺
119910120591119894119881
⊺
119861119894119898
]
Γ25119894119898
= [
ℎ119860⊺
120591119894119872
⊺
1119894119898+ ℎ119862
⊺
119910120591119894119872
⊺
119861119894119898
ℎ119860⊺
120591119894119872
⊺
2119894119898+ ℎ119862
⊺
119910120591119894119872
⊺
119861119894119898
]
Γ1015840
25119894119898= [
ℎ119860⊺
120591119894119873
⊺
1119894119898+ ℎ119862
⊺
119910120591119894119873
⊺
119861119894119898
ℎ119860⊺
120591119894119873
⊺
2119894119898+ ℎ119862
⊺
119910120591119894119873
⊺
119861119894119898
]
Γ46119894119898
= [
1198811119894119898119863
119894+ 119881
119861119894119898119863
119910119894
1198812119894119898119863
119894+ 119881
119861119894119898119863
119910119894
]
Γ56119894119898
= [
ℎ1198721119894119898119863
119894+ ℎ119872
119861119894119898119863
119910119894
ℎ1198722119894119898119863
119894+ ℎ119872
119861119894119898119863
119910119894
]
Γ1015840
56119894119898= [
ℎ1198731119894119898119863
119894+ ℎ119873
119861119894119898119863
119910119894
ℎ1198732119894119898119863
119894+ ℎ119873
119861119894119898119863
119910119894
]
(62)
Then a desired filter can be chosen with parameters as
119860119891119894119898
= 119875minus1
2119894119898119860
119891119894119898 119861
119891119894119898= 119875
minus1
2119894119898119861119891119894119898
119862119891119894119898
= 119862119891119894119898
(63)
Proof We denote that
119875119894119898= [
1198751119894119898
1198752119894119898
1198752119894119898
1198752119894119898
] (64)
The term 119875119894119898119860
119894can be rewritten as
119875119894119898119860
119894= [
1198751119894119898119860
119894+ 119875
21198941198981198611198911198941198981198621199101198941198752119894119898119860
119891119894119898
1198752119894119898119860
119894+ 119875
21198941198981198611198911198941198981198621199101198941198752119894119898119860
119891119894119898
] (65)
Similarly we have
119875119894119898119861119894= [
1198751119894119898119860
120591119894+ 119875
2119894119898119861119891119894119898119862119910120591119894
1198752119894119898119860
120591119894+ 119875
2119894119898119861119891119894119898119862119910120591119894
]
119875119894119898119863
119894= [
1198751119894119898119863
119894+ 119875
2119894119898119861119891119894119898119863
119910119894
1198752119894119898119863
119894+ 119875
2119894119898119861119891119894119898119863
119910119894
]
(66)
Define 119860119891119894119898
= 1198752119894119898119860
119891119894119898 119861
119891119894119898= 119875
2119894119898119861119891119894119898
119862119891119894119898
= 119862119891119894119898
119872
119860119894119898= 119872
2119894119898119860
119891119894119898119872
119861119894119898= 119872
2119894119898119861119891119894119898
119873119860119894119898
= 1198732119894119898119860
119891119894119898
119873119861119894119898
= 1198732119894119898119861119891119894119898
119881119860119894119898
= 1198812119894119898119860
119891119894119898 and 119881
119861119894119898= 119881
2119894119898119861119891119894119898
Therefore if (59) and (60) hold system (9) is finite-timebounded with a prescribed 119867
infinperformance index 120574 The
proof is completed
Remark 16 In many actual applications the minimum valueof 1205742min is of interest In Theorem 12 with a fixed 120582 120574min canbe obtained through the following optimization procedure
min 1205742
st (48)ndash(50) (67)
In Theorem 15 as for finite-time stability and bounded-ness once the state bound 119888
2is not ascertained theminimum
value 1198882min is of interest With a fixed 120582 define 120582
1= 1 then
the following optimization problem can be formulated to getminimum value 119888
2min
min 1205891205742
+ (1 minus 120589) 1198882
st (59)-(60) and (50)
(68)
where 120589 is weighted factor and 120589 isin [0 1]
5 Illustrative Example
Example 17 Consider that the Markovian jump system andthe delay mode switching are governed by a Markov processwith the following transition rates
Λ = [
minus07 07
09 minus09] Π = [
minus1 1
12 minus12] (69)
as well as with the following parameters
1198601= [
minus35 086
minus064 minus325] 119860
2= [
minus25 034
14 minus002]
1198601205911= [
minus08 minus13
minus07 minus22] 119860
1205912= [
minus28 05
minus08 minus14]
1198631= 119863
2= [
01
02] 119862
1199101= [04 07]
1198621199102= [05 08] 119862
1199101205911= [01 02]
1198621199101205912
= [01 01]
1198631199101= 119863
1199102= 01 119862
1199111= [01 005]
1198621199112= [02 009] 119862
1199111205911= [03 06]
1198621199111205912
= [04 08] 1198631199111= 119863
1199112= 005
(70)
Journal of Applied Mathematics 11
Then we choose 119877 = 119868 119879 = 2 1198881= 1 and 119889 = 001
throughTheorem 15 it yields that themode-dependent filtersare as follows
11986011989111
= [
minus61254 15565
minus02347 minus03763]
11986011989112
= [
minus62682 14338
minus03552 minus05327]
11986011989121
= [
minus89214 24223
minus15476 minus06234]
11986011989122
= [
minus84638 28237
minus13545 minus04322]
11986111989111
= [
51465 minus28516
minus91216 95171]
11986111989112
= [
54203 minus23121
minus93156 96332]
11986111989121
= [
46193 minus11984
minus162397 163006]
11986111989122
= [
46512 minus10311
minus161132 162312]
11986211989111
= [00294 minus00397]
11986211989112
= [00271 minus003368]
11986211989121
= [01163 minus01432]
11986211989122
= [01342 minus03672]
(71)
This paper deals with the finite-time filter design problemfor a class of Markovian jump systems particularly two dif-ferentMarkov processes are considered formodeling the ran-domness of system matrix and the state delay Then throughthe numerical example we can see that results in this paperare feasible which further verified the correctness of our the-ory Therefore the paper shorten this gap
6 Conclusions
In this paper we have examined the problems of finite-time119867
infinfiltering for a class of Markovian jump systems with dif-
ferent system modes Based on a novel approach a sufficientcondition is derived such that the closed-loop Markovianjump system is finite-time bounded and satisfies a prescribedlevel of119867
infindisturbance attenuation in a finite time interval
Finally a numerical example is also given to illustrate theeffectiveness of the proposed design approach It should benoted that one of future research topics would be to investi-gate the problems of fault detection and fault tolerant controlfor time-varying Markovian jump systems with incompleteinformation over a finite-time horizon
Acknowledgments
The authors would like to thank the associate editor and theanonymous reviewers for their detailed comments and sug-gestions This work was supported by the Fund of SichuanProvincial Key Laboratory of Signal and Information Pro-cessing Xihua University (SZJJ2009-002 and SGXZD0101-10-1) and National Basic Research Program of China(2010CB732501)
References
[1] N N Krasovskiı and E A Lidskiı ldquoAnalytical design ofcontrollers in systems with random attributesmdashI Statementof the problem method of solvingrdquo Automation and RemoteControl vol 22 pp 1021ndash1025 1961
[2] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[3] L Wu X Su and P Shi ldquoSliding mode control with bounded1198712gain performance of Markovian jump singular time-delay
systemsrdquo Automatica vol 48 no 8 pp 1929ndash1933 2012[4] H Gao Z Fei J Lam and B Du ldquoFurther results on expo-
nential estimates of Markovian jump systems with mode-de-pendent time-varying delaysrdquo IEEE Transactions on AutomaticControl vol 56 no 1 pp 223ndash229 2011
[5] Z Wu H Su and J Chu ldquo119867infinfiltering for singular Markovian
jump systems with time delayrdquo International Journal of Robustand Nonlinear Control vol 20 no 8 pp 939ndash957 2010
[6] P Tino M Cernansky and L Benuskova ldquoMarkovian architec-tural bias of recurrent neural networksrdquo IEEE Transactions onNeural Networks vol 15 no 1 pp 6ndash15 2004
[7] H Dong Z Wang D W C Ho and H Gao ldquoRobust 119867infin
filtering for Markovian jump systems with randomly occurringnonlinearities and sensor saturation the finite-horizon caserdquoIEEE Transactions on Signal Processing vol 59 no 7 pp 3048ndash3057 2011
[8] Q Ma S Xu and Y Zou ldquoStability and synchronization forMarkovian jump neural networks with partly unknown tran-sition probabilitiesrdquo Neurocomputing vol 74 pp 3404ndash33412011
[9] H Zhao S Xu and Y Zou ldquoRobust119867infinfiltering for uncertain
Markovian jump systems with mode-dependent distributeddelaysrdquo International Journal of Adaptive Control and SignalProcessing vol 24 no 1 pp 83ndash94 2010
[10] Y Zhang S Xu and B Zhang ldquoDiscrete-time fuzzy Markovianjump systems with time-varying delaysrdquo IEEE Transactions onFuzzy Systems vol 17 no 2 pp 411ndash420 2009
[11] Z Wu H Su and J Chu ldquoState estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 no 10ndash12 pp 2247ndash2254 2010
[12] Q Zhu and J Cao ldquoRobust exponential stability of markovianjump impulsive stochastic Cohen-Grossberg neural networkswithmixed time delaysrdquo IEEE Transactions onNeural Networksvol 21 no 8 pp 1314ndash1325 2010
[13] H Huang G Feng and X Chen ldquoStability and stabilizationof Markovian jump systems with time delay via new Lyapunovfunctionalsrdquo IEEE Transactions on Circuits and Systems I vol59 no 10 pp 2413ndash2421 2012
12 Journal of Applied Mathematics
[14] Q Zhu and J Cao ldquoStability of Markovian jump neural net-works with impulse control and time varying delaysrdquoNonlinearAnalysis Real World Applications vol 13 no 5 pp 2259ndash22702012
[15] M Sun J Lam S Xu andY Zou ldquoRobust exponential stabiliza-tion for Markovian jump systems with mode-dependent inputdelayrdquo Automatica vol 43 no 10 pp 1799ndash1807 2007
[16] M Mohammadian A Abolmasoumi and H Momeni ldquo119867infin
mode-independent filter design for Markovian jump geneticregulatory networks with time-varying delaysrdquo Neurocomput-ing vol 87 pp 10ndash18 2012
[17] S Boyd L El Ghaoui E Feron and V Balakrishnan LinearMatrix Inequalities in SystemandControlTheory vol 15 of SIAMStudies in Applied Mathematics SIAM Philadelphia Pa USA1994
[18] S Mou H GaoW Qiang and K Chen ldquoNew delay-dependentexponential stability for neural networks with time delayrdquo IEEETransactions on Systems Man and Cybernetics Part B vol 38no 2 pp 571ndash576 2008
[19] O M Kwon J Park S Lee and E Cha ldquoAnalysis on delay-dependent stability for neural networks with time-varyingdelaysrdquo Neurocomputing vol 103 pp 114ndash120 2013
[20] L Hu H Gao andW X Zheng ldquoNovel stability of cellular neu-ral networkswith interval time-varying delayrdquoNeuralNetworksvol 21 no 10 pp 1458ndash1463 2008
[21] H Zhang Z Liu G B Huang and ZWang ldquoNovel weighting-delay-based stability criteria for recurrent neural networks withtime-varying delayrdquo IEEE Transactions on Neural Networks vol21 no 1 pp 91ndash106 2010
[22] Z Zuo C Yang and YWang ldquoA newmethod for stability anal-ysis of recurrent neural networks with interval time-varyingdelayrdquo IEEE Transactions on Neural Networks vol 21 no 2 pp339ndash344 2010
[23] P Park JWKo andC Jeong ldquoReciprocally convex approach tostability of systems with time-varying delaysrdquo Automatica vol47 no 1 pp 235ndash238 2011
[24] D Zhang L Yu and W Zhang ldquoDelay-dependent fault detec-tion for switched linear systems with time-varying delays-theaverage dwell time approachrdquo Signal Processing vol 91 pp 832ndash840 2011
[25] Z Fei H Gao and P Shi ldquoNew results on stabilization ofMark-ovian jump systems with time delayrdquoAutomatica vol 45 no 10pp 2300ndash2306 2009
[26] H R Karimi ldquoRobust delay-dependent 119867infin
control of uncer-tain time-delay systems with mixed neutral discrete anddistributed time-delays and Markovian switching parametersrdquoIEEE Transactions on Circuits and Systems I vol 58 no 8 pp1910ndash1923 2011
[27] M Luo G Liu and S Zhong ldquoRobust fault detection ofMarkovian jump systems with different systemmodesrdquoAppliedMathematical Modelling vol 37 no 7 pp 5001ndash5012 2013
[28] B Zhang andY Li ldquoExponential1198712minus119871
infinfiltering for distributed
delay systems with Markovian jumping parametersrdquo SignalProcessing vol 93 no 1 pp 206ndash216 2013
[29] A Abolmasoumi and H Momeni ldquoRobust observer-based119867infin
control of a Markovian jump system with different delay andsystem modesrdquo International Journal of Control Automationand Systems vol 9 no 4 pp 768ndash776 2011
[30] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[31] S Ma and E K Boukas ldquoRobust 119867infin
filtering for uncertaindiscrete Markov jump singular systems with mode-dependenttime delayrdquo IET Control Theory amp Applications vol 3 no 3 pp351ndash361 2009
[32] Y ChenW Bi andW Li ldquoStability analysis for neural networkswith time-varying delay a more general delay decompositionapproachrdquo Neurocomputing vol 73 no 4ndash6 pp 853ndash857 2010
[33] Z Shu J Lam and S Xu ldquoRobust stabilization of Markoviandelay systems with delay-dependent exponential estimatesrdquoAutomatica vol 42 no 11 pp 2001ndash2008 2006
[34] X Huang W Lin and B Yang ldquoGlobal finite-time stabilizationof a class of uncertain nonlinear systemsrdquo Automatica vol 41no 5 pp 881ndash888 2005
[35] C Qian and J Li ldquoGlobal finite-time stabilization by outputfeedback for planar systems without observable linearizationrdquoIEEE Transactions on Automatic Control vol 50 no 6 pp 885ndash890 2005
[36] F Amato R Ambrosino C Cosentino and G De TommasildquoInput-output finite time stabilization of linear systemsrdquo Auto-matica vol 46 no 9 pp 1558ndash1562 2010
[37] Y Hong ldquoFinite-time stabilization and stabilizability of a classof controllable systemsrdquo Systems amp Control Letters vol 46 no4 pp 231ndash236 2002
[38] F Amato and M Ariola ldquoFinite-time control of discrete-timelinear systemsrdquo IEEETransactions onAutomatic Control vol 50no 5 pp 724ndash729 2005
[39] Z Zuo Y Liu Y Wang and H Li ldquoFinite-time stochasticstability and stabilisation of linear discrete-time Markovianjump systems with partly unknown transition probabilitiesrdquoIET Control Theory amp Applications vol 6 no 10 pp 1522ndash15262012
[40] X Luan F Liu and P Shi ldquoFinite-time filtering for non-linearstochastic systems with partially known transition jump ratesrdquoIET Control Theory amp Applications vol 4 no 5 pp 735ndash7452010
[41] F Amato M Ariola and C Cosentino ldquoFinite-time control ofdiscrete-time linear systems analysis and design conditionsrdquoAutomatica vol 46 no 5 pp 919ndash924 2010
[42] X Lin H Du and S Li ldquoFinite-time boundedness and 1198712-
gain analysis for switched delay systems with norm-boundeddisturbancerdquo Applied Mathematics and Computation vol 217no 12 pp 5982ndash5993 2011
[43] J Lin S Fei and Z Gao ldquoStabilization of discrete-time switchedsingular time-delay systems under asynchronous switchingrdquoJournal of the Franklin Institute vol 349 no 5 pp 1808ndash18272012
[44] H Song L Yu D Zhang and W-A Zhang ldquoFinite-time 119867infin
control for a class of discrete-time switched time-delay systemswith quantized feedbackrdquoCommunications in Nonlinear Scienceand Numerical Simulation vol 17 no 12 pp 4802ndash4814 2012
[45] Z Zuo H Li Y Liu and Y Wang ldquoOn finite-time stochasticstability and stabilization ofMarkovian jump systems subject topartial information on transition probabilitiesrdquoCircuits Systemsand Signal Processing 2012
[46] H Liu Y Shen and X Zhao ldquoDelay-dependent observer-based119867
infinfinite-time control for switched systems with time-varying
delayrdquoNonlinear Analysis Hybrid Systems vol 6 no 3 pp 885ndash898 2012
[47] Y Zhang C Liu and X Mu ldquoRobust finite-time119867infincontrol of
singular stochastic systems via static output feedbackrdquo AppliedMathematics and Computation vol 218 no 9 pp 5629ndash56402012
Journal of Applied Mathematics 13
[48] J Cheng H Zhu S Zhong Y Zhang and Y Zeng ldquoFinite-timeStabilization of 119867
infinfiltering for switched stochastic systemsrdquo
Circuits Systems and Signal Processing 2012[49] J Cheng H Zhu S Zhong and Y Zhang ldquoFinite-time bound-
ness of 119867infin
filtering for switching discrete-time systemsrdquo In-ternational Journal of Control Automation and Systems vol 10pp 1129ndash1135 2012
[50] Z Xiang C Qiao and M S Mahmoud ldquoFinite-time analysisand119867
infincontrol for switched stochastic systemsrdquo Journal of the
Franklin Institute vol 349 no 3 pp 915ndash927 2012
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Journal of Applied Mathematics 9
Proof We now consider the 119867infin
performance of system (9)Select the same Lyapunov-Krasovskii functional as Theo-rem 7 it yields that
pound119881 (120578119905 119903
119905 119904
119905) + 119890
⊺
(119905) 119890 (119905) minus 1205742
120596⊺
(119905) 120596 (119905) le 120585⊺
(119905) Σ119897119894119898120585 (119905)
(119897 = 1 2)
(51)
It follows from (49)-(50) that
E pound119881 (120578119905 119903
119905 119904
119905)
le E [120578119881 (120578119905 119903
119905 119904
119905)] + 120574
2
120596⊺
(119905) 120596 (119905) minus E [119890⊺
(119905) 119890 (119905)]
(52)
Multiplying the aforementioned inequality by 119890minus120578119905 onehas
E pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] le 119890
minus120578119905
[1205742
120596⊺
(119905) 120596 (119905) minus 119890⊺
(119905) 119890 (119905)]
(53)
In zero initial condition and E[119881(120578119905 119903
119905 119904
119905)] gt 0 by inte-
grating the aforementioned inequality between 0 and 119879 wecan get
int
119879
0
119890minus120578119905
[1205742
120596⊺
(119905) 120596 (119905) minus 119890⊺
(119905) 119890 (119905)] 119889119905
le Eint119879
0
pound [119890minus120578119905119881 (120578119905 119903
119905 119904
119905)] 119889119905 le 119881 (120578
0 119903
0 119904
0) = 0
(54)
Using Dynkins formula it results that
E [int119879
0
119890minus120578120592
119890⊺
(120592) 119890 (120592) 119889120592] le 1205742
E [int119879
0
119890minus120578120592
120596⊺
(120592) 120596 (120592) 119889120592]
(55)
Then it yields that
E [int119879
0
119890⊺
(120592) 119890 (120592) 119889120592] le 1205742
119890120578119879
E [int119879
0
120596⊺
(120592) 120596 (120592) 119889120592] (56)
Thus it is concluded by Definition 3 that system (9) is fi-nite-time bounded with an 119867
infinperformance 120574 This com-
pletes the proof
Remark 13 From the proof process of Theorems 7 and 12it is easy to see that neither bounding technique for cross-terms nor model transformation is involved In other wordsthe obtained result is expected to be less conservative
Remark 14 The Lyapunov asymptotic stability and finite-time stability of a class of system are independent conceptsA Lyapunov asymptotic stability system may not be finite-time stability Moreover finite-time stability system may alsonot be Lyapunov asymptotic stabilityThere exist some resultson Lyapunov stability while finite-time stability also needsour full investigation which was neglected by most previouswork
4 Finite-Time 119867infin
Filtering
Theorem 15 System (9) is finite-time bounded with respect to(1198881 119888
2 119889 119877 119879) if there exist matrices 119875
119894119898gt 0119872
119894119898 119873
119894119898 119881
119894119898
119876119897119894gt 0 (119897 = 1 2) 119876 gt 0 119883
119894119898gt 0 119883 gt 0 119884
119894119898gt 0 119884 gt 0
119860119891119894119898
119861119891119894119898
119862119891119894119898
119872119860119894119898
119872119861119894119898
119873119860119894119898
119873119861119894119898
119881119860119894119898
119881119861119894119898
119878119894119898
scalars 1198881lt 119888
2 119879 gt 0 120590
119904gt 0 (119904 = 1 2 9) 120575 gt 0 120578 gt 0
and Λ gt 0 such that for all 119894 119895 isin N and 119898 119899 isin M thefollowing inequalities hold
119875119894119898= [
1198751119894119898
1198752119894119898
1198752119894119898
1198752119894119898
] gt 0 119872119894119898= [
1198721119894119898
1198722119894119898
1198722119894119898
1198722119894119898
] (57)
119873119894119898= [
1198731119894119898
1198732119894119898
1198732119894119898
1198732119894119898
] 119881119894119898= [
1198811119894119898
1198812119894119898
1198812119894119898
1198812119894119898
] (58)
Γ1119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Γ11119894119898
Γ12119894119898
minus119878119894119898
Γ14119894119898
Γ15119894119898
Γ16119894119898
Γ17119894119898
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
Γ24119894119898
Γ25119894119898
0 [119862⊺
119911120591119894
0]
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119872⊺
119894119898Γ46119894119898
0
lowast lowast lowast lowast minusℎ119883119894119898
Γ56119894119898
0
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(59)
Γ2119894119898
=
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Γ11119894119898
Γ12119894119898
minus119878119894119898
Γ14119894119898
Γ1015840
15119894119898Γ16119894119898
Γ17119894119898
lowast Ξ22119894119898
119878119894119898
minus
119884119894119898
ℎ
Γ24119894119898
Γ1015840
251198941198980 [
119862⊺
119911120591119894
0]
lowast lowast minus1198761119894+
119884119894119898
ℎ
0 0 0 0
lowast lowast lowast Ξ44119894119898
minusℎ119873⊺
119894119898Γ1015840
461198941198980
lowast lowast lowast lowast minusℎ119883119894119898
Γ1015840
561198941198980
lowast lowast lowast lowast lowast minus1205742119868 0
lowast lowast lowast lowast lowast lowast minus119868
]
]
]
]
]
]
]
]
]
]
]
]
]
]
lt 0
(60)where
Γ11119894119898
= sum
119899isinM
120587119898119899119875119894119899+ sum
119895isinN
120582119894119895119875119895119898+ 119890
120575ℎ
1198761119894+ 119890
120575ℎ
1198762119894+ ℎ119876 +
119890120575ℎ
minus 1
120575
119883119894119898+
119890120575ℎ
minus 120575ℎ119890120575ℎ
minus 1
1205752
119883
+
119884119894119898
ℎ
+ [
1205751198751119894119898
+ 1198751119894119898119860
119894+ 119860
⊺
1198941198751119894119898
+ 119861119891119894119898119862119910119894+ 119862
⊺
119910119894119861
⊺
119891119894119898120575119875
2119894119898+ 119860
119891119894119898+ 119860
⊺
119891119894119898
1205751198752119894119898
+ 1198752119894119898119860
119894+ 119860
⊺
1198941198752119894119898
+ 119861119891119894119898119862119910119894+ 119862
⊺
119910119894119861
⊺
119891119894119898120575119875
2119894119898+ 119860
119891119894119898+ 119860
⊺
119891119894119898
]
(61)
10 Journal of Applied Mathematics
Γ12119894119898
= minus
119884119894119898
ℎ
+ 119878119894119898+ [
1198751119894119898119860
120591119894+ 119861
119891119894119898119862119910120591119894
1198752119894119898119860
120591119894+ 119861
119891119894119898119862119910120591119894
]
Γ14119894119898
= [
119860⊺
119894119881
⊺
1119894119898+ 119862
⊺
119910119894119881
⊺
119861119894119898119881
⊺
119860119894119898
119860⊺
119894119881
⊺
2119894119898+ 119862
⊺
119910119894119881
⊺
119861119894119898119881
⊺
119860119894119898
]
Γ15119894119898
= [
ℎ119860⊺
119894119872
⊺
1119894119898+ ℎ119862
⊺
119910119894119872
⊺
119861119894119898ℎ119872
⊺
119860119894119898
ℎ119860⊺
119894119872
⊺
2119894119898+ ℎ119862
⊺
119910119894119872
⊺
119861119894119898ℎ119872
⊺
119860119894119898
]
Γ1015840
15119894119898= [
ℎ119860⊺
119894119873
⊺
1119894119898+ ℎ119862
⊺
119910119894119873
⊺
119861119894119898ℎ119873
⊺
119860119894119898
ℎ119860⊺
119894119873
⊺
2119894119898+ ℎ119862
⊺
119910119894119873
⊺
119861119894119898ℎ119873
⊺
119860119894119898
]
Γ16119894119898
= [
1198751119894119898119863
119894+ 119861
119891119894119898119863
119910119894
1198752119894119898119863
119894+ 119861
119891119894119898119863
119910119894
]
Γ17119894119898
= [
119862⊺
119911119894
minus119862119891119894119898
]
Γ24119894119898
= [
119860⊺
120591119894119881
⊺
1119894119898+ 119862
⊺
119910120591119894119881
⊺
119861119894119898
119860⊺
120591119894119881
⊺
2119894119898+ 119862
⊺
119910120591119894119881
⊺
119861119894119898
]
Γ25119894119898
= [
ℎ119860⊺
120591119894119872
⊺
1119894119898+ ℎ119862
⊺
119910120591119894119872
⊺
119861119894119898
ℎ119860⊺
120591119894119872
⊺
2119894119898+ ℎ119862
⊺
119910120591119894119872
⊺
119861119894119898
]
Γ1015840
25119894119898= [
ℎ119860⊺
120591119894119873
⊺
1119894119898+ ℎ119862
⊺
119910120591119894119873
⊺
119861119894119898
ℎ119860⊺
120591119894119873
⊺
2119894119898+ ℎ119862
⊺
119910120591119894119873
⊺
119861119894119898
]
Γ46119894119898
= [
1198811119894119898119863
119894+ 119881
119861119894119898119863
119910119894
1198812119894119898119863
119894+ 119881
119861119894119898119863
119910119894
]
Γ56119894119898
= [
ℎ1198721119894119898119863
119894+ ℎ119872
119861119894119898119863
119910119894
ℎ1198722119894119898119863
119894+ ℎ119872
119861119894119898119863
119910119894
]
Γ1015840
56119894119898= [
ℎ1198731119894119898119863
119894+ ℎ119873
119861119894119898119863
119910119894
ℎ1198732119894119898119863
119894+ ℎ119873
119861119894119898119863
119910119894
]
(62)
Then a desired filter can be chosen with parameters as
119860119891119894119898
= 119875minus1
2119894119898119860
119891119894119898 119861
119891119894119898= 119875
minus1
2119894119898119861119891119894119898
119862119891119894119898
= 119862119891119894119898
(63)
Proof We denote that
119875119894119898= [
1198751119894119898
1198752119894119898
1198752119894119898
1198752119894119898
] (64)
The term 119875119894119898119860
119894can be rewritten as
119875119894119898119860
119894= [
1198751119894119898119860
119894+ 119875
21198941198981198611198911198941198981198621199101198941198752119894119898119860
119891119894119898
1198752119894119898119860
119894+ 119875
21198941198981198611198911198941198981198621199101198941198752119894119898119860
119891119894119898
] (65)
Similarly we have
119875119894119898119861119894= [
1198751119894119898119860
120591119894+ 119875
2119894119898119861119891119894119898119862119910120591119894
1198752119894119898119860
120591119894+ 119875
2119894119898119861119891119894119898119862119910120591119894
]
119875119894119898119863
119894= [
1198751119894119898119863
119894+ 119875
2119894119898119861119891119894119898119863
119910119894
1198752119894119898119863
119894+ 119875
2119894119898119861119891119894119898119863
119910119894
]
(66)
Define 119860119891119894119898
= 1198752119894119898119860
119891119894119898 119861
119891119894119898= 119875
2119894119898119861119891119894119898
119862119891119894119898
= 119862119891119894119898
119872
119860119894119898= 119872
2119894119898119860
119891119894119898119872
119861119894119898= 119872
2119894119898119861119891119894119898
119873119860119894119898
= 1198732119894119898119860
119891119894119898
119873119861119894119898
= 1198732119894119898119861119891119894119898
119881119860119894119898
= 1198812119894119898119860
119891119894119898 and 119881
119861119894119898= 119881
2119894119898119861119891119894119898
Therefore if (59) and (60) hold system (9) is finite-timebounded with a prescribed 119867
infinperformance index 120574 The
proof is completed
Remark 16 In many actual applications the minimum valueof 1205742min is of interest In Theorem 12 with a fixed 120582 120574min canbe obtained through the following optimization procedure
min 1205742
st (48)ndash(50) (67)
In Theorem 15 as for finite-time stability and bounded-ness once the state bound 119888
2is not ascertained theminimum
value 1198882min is of interest With a fixed 120582 define 120582
1= 1 then
the following optimization problem can be formulated to getminimum value 119888
2min
min 1205891205742
+ (1 minus 120589) 1198882
st (59)-(60) and (50)
(68)
where 120589 is weighted factor and 120589 isin [0 1]
5 Illustrative Example
Example 17 Consider that the Markovian jump system andthe delay mode switching are governed by a Markov processwith the following transition rates
Λ = [
minus07 07
09 minus09] Π = [
minus1 1
12 minus12] (69)
as well as with the following parameters
1198601= [
minus35 086
minus064 minus325] 119860
2= [
minus25 034
14 minus002]
1198601205911= [
minus08 minus13
minus07 minus22] 119860
1205912= [
minus28 05
minus08 minus14]
1198631= 119863
2= [
01
02] 119862
1199101= [04 07]
1198621199102= [05 08] 119862
1199101205911= [01 02]
1198621199101205912
= [01 01]
1198631199101= 119863
1199102= 01 119862
1199111= [01 005]
1198621199112= [02 009] 119862
1199111205911= [03 06]
1198621199111205912
= [04 08] 1198631199111= 119863
1199112= 005
(70)
Journal of Applied Mathematics 11
Then we choose 119877 = 119868 119879 = 2 1198881= 1 and 119889 = 001
throughTheorem 15 it yields that themode-dependent filtersare as follows
11986011989111
= [
minus61254 15565
minus02347 minus03763]
11986011989112
= [
minus62682 14338
minus03552 minus05327]
11986011989121
= [
minus89214 24223
minus15476 minus06234]
11986011989122
= [
minus84638 28237
minus13545 minus04322]
11986111989111
= [
51465 minus28516
minus91216 95171]
11986111989112
= [
54203 minus23121
minus93156 96332]
11986111989121
= [
46193 minus11984
minus162397 163006]
11986111989122
= [
46512 minus10311
minus161132 162312]
11986211989111
= [00294 minus00397]
11986211989112
= [00271 minus003368]
11986211989121
= [01163 minus01432]
11986211989122
= [01342 minus03672]
(71)
This paper deals with the finite-time filter design problemfor a class of Markovian jump systems particularly two dif-ferentMarkov processes are considered formodeling the ran-domness of system matrix and the state delay Then throughthe numerical example we can see that results in this paperare feasible which further verified the correctness of our the-ory Therefore the paper shorten this gap
6 Conclusions
In this paper we have examined the problems of finite-time119867
infinfiltering for a class of Markovian jump systems with dif-
ferent system modes Based on a novel approach a sufficientcondition is derived such that the closed-loop Markovianjump system is finite-time bounded and satisfies a prescribedlevel of119867
infindisturbance attenuation in a finite time interval
Finally a numerical example is also given to illustrate theeffectiveness of the proposed design approach It should benoted that one of future research topics would be to investi-gate the problems of fault detection and fault tolerant controlfor time-varying Markovian jump systems with incompleteinformation over a finite-time horizon
Acknowledgments
The authors would like to thank the associate editor and theanonymous reviewers for their detailed comments and sug-gestions This work was supported by the Fund of SichuanProvincial Key Laboratory of Signal and Information Pro-cessing Xihua University (SZJJ2009-002 and SGXZD0101-10-1) and National Basic Research Program of China(2010CB732501)
References
[1] N N Krasovskiı and E A Lidskiı ldquoAnalytical design ofcontrollers in systems with random attributesmdashI Statementof the problem method of solvingrdquo Automation and RemoteControl vol 22 pp 1021ndash1025 1961
[2] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[3] L Wu X Su and P Shi ldquoSliding mode control with bounded1198712gain performance of Markovian jump singular time-delay
systemsrdquo Automatica vol 48 no 8 pp 1929ndash1933 2012[4] H Gao Z Fei J Lam and B Du ldquoFurther results on expo-
nential estimates of Markovian jump systems with mode-de-pendent time-varying delaysrdquo IEEE Transactions on AutomaticControl vol 56 no 1 pp 223ndash229 2011
[5] Z Wu H Su and J Chu ldquo119867infinfiltering for singular Markovian
jump systems with time delayrdquo International Journal of Robustand Nonlinear Control vol 20 no 8 pp 939ndash957 2010
[6] P Tino M Cernansky and L Benuskova ldquoMarkovian architec-tural bias of recurrent neural networksrdquo IEEE Transactions onNeural Networks vol 15 no 1 pp 6ndash15 2004
[7] H Dong Z Wang D W C Ho and H Gao ldquoRobust 119867infin
filtering for Markovian jump systems with randomly occurringnonlinearities and sensor saturation the finite-horizon caserdquoIEEE Transactions on Signal Processing vol 59 no 7 pp 3048ndash3057 2011
[8] Q Ma S Xu and Y Zou ldquoStability and synchronization forMarkovian jump neural networks with partly unknown tran-sition probabilitiesrdquo Neurocomputing vol 74 pp 3404ndash33412011
[9] H Zhao S Xu and Y Zou ldquoRobust119867infinfiltering for uncertain
Markovian jump systems with mode-dependent distributeddelaysrdquo International Journal of Adaptive Control and SignalProcessing vol 24 no 1 pp 83ndash94 2010
[10] Y Zhang S Xu and B Zhang ldquoDiscrete-time fuzzy Markovianjump systems with time-varying delaysrdquo IEEE Transactions onFuzzy Systems vol 17 no 2 pp 411ndash420 2009
[11] Z Wu H Su and J Chu ldquoState estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 no 10ndash12 pp 2247ndash2254 2010
[12] Q Zhu and J Cao ldquoRobust exponential stability of markovianjump impulsive stochastic Cohen-Grossberg neural networkswithmixed time delaysrdquo IEEE Transactions onNeural Networksvol 21 no 8 pp 1314ndash1325 2010
[13] H Huang G Feng and X Chen ldquoStability and stabilizationof Markovian jump systems with time delay via new Lyapunovfunctionalsrdquo IEEE Transactions on Circuits and Systems I vol59 no 10 pp 2413ndash2421 2012
12 Journal of Applied Mathematics
[14] Q Zhu and J Cao ldquoStability of Markovian jump neural net-works with impulse control and time varying delaysrdquoNonlinearAnalysis Real World Applications vol 13 no 5 pp 2259ndash22702012
[15] M Sun J Lam S Xu andY Zou ldquoRobust exponential stabiliza-tion for Markovian jump systems with mode-dependent inputdelayrdquo Automatica vol 43 no 10 pp 1799ndash1807 2007
[16] M Mohammadian A Abolmasoumi and H Momeni ldquo119867infin
mode-independent filter design for Markovian jump geneticregulatory networks with time-varying delaysrdquo Neurocomput-ing vol 87 pp 10ndash18 2012
[17] S Boyd L El Ghaoui E Feron and V Balakrishnan LinearMatrix Inequalities in SystemandControlTheory vol 15 of SIAMStudies in Applied Mathematics SIAM Philadelphia Pa USA1994
[18] S Mou H GaoW Qiang and K Chen ldquoNew delay-dependentexponential stability for neural networks with time delayrdquo IEEETransactions on Systems Man and Cybernetics Part B vol 38no 2 pp 571ndash576 2008
[19] O M Kwon J Park S Lee and E Cha ldquoAnalysis on delay-dependent stability for neural networks with time-varyingdelaysrdquo Neurocomputing vol 103 pp 114ndash120 2013
[20] L Hu H Gao andW X Zheng ldquoNovel stability of cellular neu-ral networkswith interval time-varying delayrdquoNeuralNetworksvol 21 no 10 pp 1458ndash1463 2008
[21] H Zhang Z Liu G B Huang and ZWang ldquoNovel weighting-delay-based stability criteria for recurrent neural networks withtime-varying delayrdquo IEEE Transactions on Neural Networks vol21 no 1 pp 91ndash106 2010
[22] Z Zuo C Yang and YWang ldquoA newmethod for stability anal-ysis of recurrent neural networks with interval time-varyingdelayrdquo IEEE Transactions on Neural Networks vol 21 no 2 pp339ndash344 2010
[23] P Park JWKo andC Jeong ldquoReciprocally convex approach tostability of systems with time-varying delaysrdquo Automatica vol47 no 1 pp 235ndash238 2011
[24] D Zhang L Yu and W Zhang ldquoDelay-dependent fault detec-tion for switched linear systems with time-varying delays-theaverage dwell time approachrdquo Signal Processing vol 91 pp 832ndash840 2011
[25] Z Fei H Gao and P Shi ldquoNew results on stabilization ofMark-ovian jump systems with time delayrdquoAutomatica vol 45 no 10pp 2300ndash2306 2009
[26] H R Karimi ldquoRobust delay-dependent 119867infin
control of uncer-tain time-delay systems with mixed neutral discrete anddistributed time-delays and Markovian switching parametersrdquoIEEE Transactions on Circuits and Systems I vol 58 no 8 pp1910ndash1923 2011
[27] M Luo G Liu and S Zhong ldquoRobust fault detection ofMarkovian jump systems with different systemmodesrdquoAppliedMathematical Modelling vol 37 no 7 pp 5001ndash5012 2013
[28] B Zhang andY Li ldquoExponential1198712minus119871
infinfiltering for distributed
delay systems with Markovian jumping parametersrdquo SignalProcessing vol 93 no 1 pp 206ndash216 2013
[29] A Abolmasoumi and H Momeni ldquoRobust observer-based119867infin
control of a Markovian jump system with different delay andsystem modesrdquo International Journal of Control Automationand Systems vol 9 no 4 pp 768ndash776 2011
[30] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[31] S Ma and E K Boukas ldquoRobust 119867infin
filtering for uncertaindiscrete Markov jump singular systems with mode-dependenttime delayrdquo IET Control Theory amp Applications vol 3 no 3 pp351ndash361 2009
[32] Y ChenW Bi andW Li ldquoStability analysis for neural networkswith time-varying delay a more general delay decompositionapproachrdquo Neurocomputing vol 73 no 4ndash6 pp 853ndash857 2010
[33] Z Shu J Lam and S Xu ldquoRobust stabilization of Markoviandelay systems with delay-dependent exponential estimatesrdquoAutomatica vol 42 no 11 pp 2001ndash2008 2006
[34] X Huang W Lin and B Yang ldquoGlobal finite-time stabilizationof a class of uncertain nonlinear systemsrdquo Automatica vol 41no 5 pp 881ndash888 2005
[35] C Qian and J Li ldquoGlobal finite-time stabilization by outputfeedback for planar systems without observable linearizationrdquoIEEE Transactions on Automatic Control vol 50 no 6 pp 885ndash890 2005
[36] F Amato R Ambrosino C Cosentino and G De TommasildquoInput-output finite time stabilization of linear systemsrdquo Auto-matica vol 46 no 9 pp 1558ndash1562 2010
[37] Y Hong ldquoFinite-time stabilization and stabilizability of a classof controllable systemsrdquo Systems amp Control Letters vol 46 no4 pp 231ndash236 2002
[38] F Amato and M Ariola ldquoFinite-time control of discrete-timelinear systemsrdquo IEEETransactions onAutomatic Control vol 50no 5 pp 724ndash729 2005
[39] Z Zuo Y Liu Y Wang and H Li ldquoFinite-time stochasticstability and stabilisation of linear discrete-time Markovianjump systems with partly unknown transition probabilitiesrdquoIET Control Theory amp Applications vol 6 no 10 pp 1522ndash15262012
[40] X Luan F Liu and P Shi ldquoFinite-time filtering for non-linearstochastic systems with partially known transition jump ratesrdquoIET Control Theory amp Applications vol 4 no 5 pp 735ndash7452010
[41] F Amato M Ariola and C Cosentino ldquoFinite-time control ofdiscrete-time linear systems analysis and design conditionsrdquoAutomatica vol 46 no 5 pp 919ndash924 2010
[42] X Lin H Du and S Li ldquoFinite-time boundedness and 1198712-
gain analysis for switched delay systems with norm-boundeddisturbancerdquo Applied Mathematics and Computation vol 217no 12 pp 5982ndash5993 2011
[43] J Lin S Fei and Z Gao ldquoStabilization of discrete-time switchedsingular time-delay systems under asynchronous switchingrdquoJournal of the Franklin Institute vol 349 no 5 pp 1808ndash18272012
[44] H Song L Yu D Zhang and W-A Zhang ldquoFinite-time 119867infin
control for a class of discrete-time switched time-delay systemswith quantized feedbackrdquoCommunications in Nonlinear Scienceand Numerical Simulation vol 17 no 12 pp 4802ndash4814 2012
[45] Z Zuo H Li Y Liu and Y Wang ldquoOn finite-time stochasticstability and stabilization ofMarkovian jump systems subject topartial information on transition probabilitiesrdquoCircuits Systemsand Signal Processing 2012
[46] H Liu Y Shen and X Zhao ldquoDelay-dependent observer-based119867
infinfinite-time control for switched systems with time-varying
delayrdquoNonlinear Analysis Hybrid Systems vol 6 no 3 pp 885ndash898 2012
[47] Y Zhang C Liu and X Mu ldquoRobust finite-time119867infincontrol of
singular stochastic systems via static output feedbackrdquo AppliedMathematics and Computation vol 218 no 9 pp 5629ndash56402012
Journal of Applied Mathematics 13
[48] J Cheng H Zhu S Zhong Y Zhang and Y Zeng ldquoFinite-timeStabilization of 119867
infinfiltering for switched stochastic systemsrdquo
Circuits Systems and Signal Processing 2012[49] J Cheng H Zhu S Zhong and Y Zhang ldquoFinite-time bound-
ness of 119867infin
filtering for switching discrete-time systemsrdquo In-ternational Journal of Control Automation and Systems vol 10pp 1129ndash1135 2012
[50] Z Xiang C Qiao and M S Mahmoud ldquoFinite-time analysisand119867
infincontrol for switched stochastic systemsrdquo Journal of the
Franklin Institute vol 349 no 3 pp 915ndash927 2012
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
10 Journal of Applied Mathematics
Γ12119894119898
= minus
119884119894119898
ℎ
+ 119878119894119898+ [
1198751119894119898119860
120591119894+ 119861
119891119894119898119862119910120591119894
1198752119894119898119860
120591119894+ 119861
119891119894119898119862119910120591119894
]
Γ14119894119898
= [
119860⊺
119894119881
⊺
1119894119898+ 119862
⊺
119910119894119881
⊺
119861119894119898119881
⊺
119860119894119898
119860⊺
119894119881
⊺
2119894119898+ 119862
⊺
119910119894119881
⊺
119861119894119898119881
⊺
119860119894119898
]
Γ15119894119898
= [
ℎ119860⊺
119894119872
⊺
1119894119898+ ℎ119862
⊺
119910119894119872
⊺
119861119894119898ℎ119872
⊺
119860119894119898
ℎ119860⊺
119894119872
⊺
2119894119898+ ℎ119862
⊺
119910119894119872
⊺
119861119894119898ℎ119872
⊺
119860119894119898
]
Γ1015840
15119894119898= [
ℎ119860⊺
119894119873
⊺
1119894119898+ ℎ119862
⊺
119910119894119873
⊺
119861119894119898ℎ119873
⊺
119860119894119898
ℎ119860⊺
119894119873
⊺
2119894119898+ ℎ119862
⊺
119910119894119873
⊺
119861119894119898ℎ119873
⊺
119860119894119898
]
Γ16119894119898
= [
1198751119894119898119863
119894+ 119861
119891119894119898119863
119910119894
1198752119894119898119863
119894+ 119861
119891119894119898119863
119910119894
]
Γ17119894119898
= [
119862⊺
119911119894
minus119862119891119894119898
]
Γ24119894119898
= [
119860⊺
120591119894119881
⊺
1119894119898+ 119862
⊺
119910120591119894119881
⊺
119861119894119898
119860⊺
120591119894119881
⊺
2119894119898+ 119862
⊺
119910120591119894119881
⊺
119861119894119898
]
Γ25119894119898
= [
ℎ119860⊺
120591119894119872
⊺
1119894119898+ ℎ119862
⊺
119910120591119894119872
⊺
119861119894119898
ℎ119860⊺
120591119894119872
⊺
2119894119898+ ℎ119862
⊺
119910120591119894119872
⊺
119861119894119898
]
Γ1015840
25119894119898= [
ℎ119860⊺
120591119894119873
⊺
1119894119898+ ℎ119862
⊺
119910120591119894119873
⊺
119861119894119898
ℎ119860⊺
120591119894119873
⊺
2119894119898+ ℎ119862
⊺
119910120591119894119873
⊺
119861119894119898
]
Γ46119894119898
= [
1198811119894119898119863
119894+ 119881
119861119894119898119863
119910119894
1198812119894119898119863
119894+ 119881
119861119894119898119863
119910119894
]
Γ56119894119898
= [
ℎ1198721119894119898119863
119894+ ℎ119872
119861119894119898119863
119910119894
ℎ1198722119894119898119863
119894+ ℎ119872
119861119894119898119863
119910119894
]
Γ1015840
56119894119898= [
ℎ1198731119894119898119863
119894+ ℎ119873
119861119894119898119863
119910119894
ℎ1198732119894119898119863
119894+ ℎ119873
119861119894119898119863
119910119894
]
(62)
Then a desired filter can be chosen with parameters as
119860119891119894119898
= 119875minus1
2119894119898119860
119891119894119898 119861
119891119894119898= 119875
minus1
2119894119898119861119891119894119898
119862119891119894119898
= 119862119891119894119898
(63)
Proof We denote that
119875119894119898= [
1198751119894119898
1198752119894119898
1198752119894119898
1198752119894119898
] (64)
The term 119875119894119898119860
119894can be rewritten as
119875119894119898119860
119894= [
1198751119894119898119860
119894+ 119875
21198941198981198611198911198941198981198621199101198941198752119894119898119860
119891119894119898
1198752119894119898119860
119894+ 119875
21198941198981198611198911198941198981198621199101198941198752119894119898119860
119891119894119898
] (65)
Similarly we have
119875119894119898119861119894= [
1198751119894119898119860
120591119894+ 119875
2119894119898119861119891119894119898119862119910120591119894
1198752119894119898119860
120591119894+ 119875
2119894119898119861119891119894119898119862119910120591119894
]
119875119894119898119863
119894= [
1198751119894119898119863
119894+ 119875
2119894119898119861119891119894119898119863
119910119894
1198752119894119898119863
119894+ 119875
2119894119898119861119891119894119898119863
119910119894
]
(66)
Define 119860119891119894119898
= 1198752119894119898119860
119891119894119898 119861
119891119894119898= 119875
2119894119898119861119891119894119898
119862119891119894119898
= 119862119891119894119898
119872
119860119894119898= 119872
2119894119898119860
119891119894119898119872
119861119894119898= 119872
2119894119898119861119891119894119898
119873119860119894119898
= 1198732119894119898119860
119891119894119898
119873119861119894119898
= 1198732119894119898119861119891119894119898
119881119860119894119898
= 1198812119894119898119860
119891119894119898 and 119881
119861119894119898= 119881
2119894119898119861119891119894119898
Therefore if (59) and (60) hold system (9) is finite-timebounded with a prescribed 119867
infinperformance index 120574 The
proof is completed
Remark 16 In many actual applications the minimum valueof 1205742min is of interest In Theorem 12 with a fixed 120582 120574min canbe obtained through the following optimization procedure
min 1205742
st (48)ndash(50) (67)
In Theorem 15 as for finite-time stability and bounded-ness once the state bound 119888
2is not ascertained theminimum
value 1198882min is of interest With a fixed 120582 define 120582
1= 1 then
the following optimization problem can be formulated to getminimum value 119888
2min
min 1205891205742
+ (1 minus 120589) 1198882
st (59)-(60) and (50)
(68)
where 120589 is weighted factor and 120589 isin [0 1]
5 Illustrative Example
Example 17 Consider that the Markovian jump system andthe delay mode switching are governed by a Markov processwith the following transition rates
Λ = [
minus07 07
09 minus09] Π = [
minus1 1
12 minus12] (69)
as well as with the following parameters
1198601= [
minus35 086
minus064 minus325] 119860
2= [
minus25 034
14 minus002]
1198601205911= [
minus08 minus13
minus07 minus22] 119860
1205912= [
minus28 05
minus08 minus14]
1198631= 119863
2= [
01
02] 119862
1199101= [04 07]
1198621199102= [05 08] 119862
1199101205911= [01 02]
1198621199101205912
= [01 01]
1198631199101= 119863
1199102= 01 119862
1199111= [01 005]
1198621199112= [02 009] 119862
1199111205911= [03 06]
1198621199111205912
= [04 08] 1198631199111= 119863
1199112= 005
(70)
Journal of Applied Mathematics 11
Then we choose 119877 = 119868 119879 = 2 1198881= 1 and 119889 = 001
throughTheorem 15 it yields that themode-dependent filtersare as follows
11986011989111
= [
minus61254 15565
minus02347 minus03763]
11986011989112
= [
minus62682 14338
minus03552 minus05327]
11986011989121
= [
minus89214 24223
minus15476 minus06234]
11986011989122
= [
minus84638 28237
minus13545 minus04322]
11986111989111
= [
51465 minus28516
minus91216 95171]
11986111989112
= [
54203 minus23121
minus93156 96332]
11986111989121
= [
46193 minus11984
minus162397 163006]
11986111989122
= [
46512 minus10311
minus161132 162312]
11986211989111
= [00294 minus00397]
11986211989112
= [00271 minus003368]
11986211989121
= [01163 minus01432]
11986211989122
= [01342 minus03672]
(71)
This paper deals with the finite-time filter design problemfor a class of Markovian jump systems particularly two dif-ferentMarkov processes are considered formodeling the ran-domness of system matrix and the state delay Then throughthe numerical example we can see that results in this paperare feasible which further verified the correctness of our the-ory Therefore the paper shorten this gap
6 Conclusions
In this paper we have examined the problems of finite-time119867
infinfiltering for a class of Markovian jump systems with dif-
ferent system modes Based on a novel approach a sufficientcondition is derived such that the closed-loop Markovianjump system is finite-time bounded and satisfies a prescribedlevel of119867
infindisturbance attenuation in a finite time interval
Finally a numerical example is also given to illustrate theeffectiveness of the proposed design approach It should benoted that one of future research topics would be to investi-gate the problems of fault detection and fault tolerant controlfor time-varying Markovian jump systems with incompleteinformation over a finite-time horizon
Acknowledgments
The authors would like to thank the associate editor and theanonymous reviewers for their detailed comments and sug-gestions This work was supported by the Fund of SichuanProvincial Key Laboratory of Signal and Information Pro-cessing Xihua University (SZJJ2009-002 and SGXZD0101-10-1) and National Basic Research Program of China(2010CB732501)
References
[1] N N Krasovskiı and E A Lidskiı ldquoAnalytical design ofcontrollers in systems with random attributesmdashI Statementof the problem method of solvingrdquo Automation and RemoteControl vol 22 pp 1021ndash1025 1961
[2] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[3] L Wu X Su and P Shi ldquoSliding mode control with bounded1198712gain performance of Markovian jump singular time-delay
systemsrdquo Automatica vol 48 no 8 pp 1929ndash1933 2012[4] H Gao Z Fei J Lam and B Du ldquoFurther results on expo-
nential estimates of Markovian jump systems with mode-de-pendent time-varying delaysrdquo IEEE Transactions on AutomaticControl vol 56 no 1 pp 223ndash229 2011
[5] Z Wu H Su and J Chu ldquo119867infinfiltering for singular Markovian
jump systems with time delayrdquo International Journal of Robustand Nonlinear Control vol 20 no 8 pp 939ndash957 2010
[6] P Tino M Cernansky and L Benuskova ldquoMarkovian architec-tural bias of recurrent neural networksrdquo IEEE Transactions onNeural Networks vol 15 no 1 pp 6ndash15 2004
[7] H Dong Z Wang D W C Ho and H Gao ldquoRobust 119867infin
filtering for Markovian jump systems with randomly occurringnonlinearities and sensor saturation the finite-horizon caserdquoIEEE Transactions on Signal Processing vol 59 no 7 pp 3048ndash3057 2011
[8] Q Ma S Xu and Y Zou ldquoStability and synchronization forMarkovian jump neural networks with partly unknown tran-sition probabilitiesrdquo Neurocomputing vol 74 pp 3404ndash33412011
[9] H Zhao S Xu and Y Zou ldquoRobust119867infinfiltering for uncertain
Markovian jump systems with mode-dependent distributeddelaysrdquo International Journal of Adaptive Control and SignalProcessing vol 24 no 1 pp 83ndash94 2010
[10] Y Zhang S Xu and B Zhang ldquoDiscrete-time fuzzy Markovianjump systems with time-varying delaysrdquo IEEE Transactions onFuzzy Systems vol 17 no 2 pp 411ndash420 2009
[11] Z Wu H Su and J Chu ldquoState estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 no 10ndash12 pp 2247ndash2254 2010
[12] Q Zhu and J Cao ldquoRobust exponential stability of markovianjump impulsive stochastic Cohen-Grossberg neural networkswithmixed time delaysrdquo IEEE Transactions onNeural Networksvol 21 no 8 pp 1314ndash1325 2010
[13] H Huang G Feng and X Chen ldquoStability and stabilizationof Markovian jump systems with time delay via new Lyapunovfunctionalsrdquo IEEE Transactions on Circuits and Systems I vol59 no 10 pp 2413ndash2421 2012
12 Journal of Applied Mathematics
[14] Q Zhu and J Cao ldquoStability of Markovian jump neural net-works with impulse control and time varying delaysrdquoNonlinearAnalysis Real World Applications vol 13 no 5 pp 2259ndash22702012
[15] M Sun J Lam S Xu andY Zou ldquoRobust exponential stabiliza-tion for Markovian jump systems with mode-dependent inputdelayrdquo Automatica vol 43 no 10 pp 1799ndash1807 2007
[16] M Mohammadian A Abolmasoumi and H Momeni ldquo119867infin
mode-independent filter design for Markovian jump geneticregulatory networks with time-varying delaysrdquo Neurocomput-ing vol 87 pp 10ndash18 2012
[17] S Boyd L El Ghaoui E Feron and V Balakrishnan LinearMatrix Inequalities in SystemandControlTheory vol 15 of SIAMStudies in Applied Mathematics SIAM Philadelphia Pa USA1994
[18] S Mou H GaoW Qiang and K Chen ldquoNew delay-dependentexponential stability for neural networks with time delayrdquo IEEETransactions on Systems Man and Cybernetics Part B vol 38no 2 pp 571ndash576 2008
[19] O M Kwon J Park S Lee and E Cha ldquoAnalysis on delay-dependent stability for neural networks with time-varyingdelaysrdquo Neurocomputing vol 103 pp 114ndash120 2013
[20] L Hu H Gao andW X Zheng ldquoNovel stability of cellular neu-ral networkswith interval time-varying delayrdquoNeuralNetworksvol 21 no 10 pp 1458ndash1463 2008
[21] H Zhang Z Liu G B Huang and ZWang ldquoNovel weighting-delay-based stability criteria for recurrent neural networks withtime-varying delayrdquo IEEE Transactions on Neural Networks vol21 no 1 pp 91ndash106 2010
[22] Z Zuo C Yang and YWang ldquoA newmethod for stability anal-ysis of recurrent neural networks with interval time-varyingdelayrdquo IEEE Transactions on Neural Networks vol 21 no 2 pp339ndash344 2010
[23] P Park JWKo andC Jeong ldquoReciprocally convex approach tostability of systems with time-varying delaysrdquo Automatica vol47 no 1 pp 235ndash238 2011
[24] D Zhang L Yu and W Zhang ldquoDelay-dependent fault detec-tion for switched linear systems with time-varying delays-theaverage dwell time approachrdquo Signal Processing vol 91 pp 832ndash840 2011
[25] Z Fei H Gao and P Shi ldquoNew results on stabilization ofMark-ovian jump systems with time delayrdquoAutomatica vol 45 no 10pp 2300ndash2306 2009
[26] H R Karimi ldquoRobust delay-dependent 119867infin
control of uncer-tain time-delay systems with mixed neutral discrete anddistributed time-delays and Markovian switching parametersrdquoIEEE Transactions on Circuits and Systems I vol 58 no 8 pp1910ndash1923 2011
[27] M Luo G Liu and S Zhong ldquoRobust fault detection ofMarkovian jump systems with different systemmodesrdquoAppliedMathematical Modelling vol 37 no 7 pp 5001ndash5012 2013
[28] B Zhang andY Li ldquoExponential1198712minus119871
infinfiltering for distributed
delay systems with Markovian jumping parametersrdquo SignalProcessing vol 93 no 1 pp 206ndash216 2013
[29] A Abolmasoumi and H Momeni ldquoRobust observer-based119867infin
control of a Markovian jump system with different delay andsystem modesrdquo International Journal of Control Automationand Systems vol 9 no 4 pp 768ndash776 2011
[30] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[31] S Ma and E K Boukas ldquoRobust 119867infin
filtering for uncertaindiscrete Markov jump singular systems with mode-dependenttime delayrdquo IET Control Theory amp Applications vol 3 no 3 pp351ndash361 2009
[32] Y ChenW Bi andW Li ldquoStability analysis for neural networkswith time-varying delay a more general delay decompositionapproachrdquo Neurocomputing vol 73 no 4ndash6 pp 853ndash857 2010
[33] Z Shu J Lam and S Xu ldquoRobust stabilization of Markoviandelay systems with delay-dependent exponential estimatesrdquoAutomatica vol 42 no 11 pp 2001ndash2008 2006
[34] X Huang W Lin and B Yang ldquoGlobal finite-time stabilizationof a class of uncertain nonlinear systemsrdquo Automatica vol 41no 5 pp 881ndash888 2005
[35] C Qian and J Li ldquoGlobal finite-time stabilization by outputfeedback for planar systems without observable linearizationrdquoIEEE Transactions on Automatic Control vol 50 no 6 pp 885ndash890 2005
[36] F Amato R Ambrosino C Cosentino and G De TommasildquoInput-output finite time stabilization of linear systemsrdquo Auto-matica vol 46 no 9 pp 1558ndash1562 2010
[37] Y Hong ldquoFinite-time stabilization and stabilizability of a classof controllable systemsrdquo Systems amp Control Letters vol 46 no4 pp 231ndash236 2002
[38] F Amato and M Ariola ldquoFinite-time control of discrete-timelinear systemsrdquo IEEETransactions onAutomatic Control vol 50no 5 pp 724ndash729 2005
[39] Z Zuo Y Liu Y Wang and H Li ldquoFinite-time stochasticstability and stabilisation of linear discrete-time Markovianjump systems with partly unknown transition probabilitiesrdquoIET Control Theory amp Applications vol 6 no 10 pp 1522ndash15262012
[40] X Luan F Liu and P Shi ldquoFinite-time filtering for non-linearstochastic systems with partially known transition jump ratesrdquoIET Control Theory amp Applications vol 4 no 5 pp 735ndash7452010
[41] F Amato M Ariola and C Cosentino ldquoFinite-time control ofdiscrete-time linear systems analysis and design conditionsrdquoAutomatica vol 46 no 5 pp 919ndash924 2010
[42] X Lin H Du and S Li ldquoFinite-time boundedness and 1198712-
gain analysis for switched delay systems with norm-boundeddisturbancerdquo Applied Mathematics and Computation vol 217no 12 pp 5982ndash5993 2011
[43] J Lin S Fei and Z Gao ldquoStabilization of discrete-time switchedsingular time-delay systems under asynchronous switchingrdquoJournal of the Franklin Institute vol 349 no 5 pp 1808ndash18272012
[44] H Song L Yu D Zhang and W-A Zhang ldquoFinite-time 119867infin
control for a class of discrete-time switched time-delay systemswith quantized feedbackrdquoCommunications in Nonlinear Scienceand Numerical Simulation vol 17 no 12 pp 4802ndash4814 2012
[45] Z Zuo H Li Y Liu and Y Wang ldquoOn finite-time stochasticstability and stabilization ofMarkovian jump systems subject topartial information on transition probabilitiesrdquoCircuits Systemsand Signal Processing 2012
[46] H Liu Y Shen and X Zhao ldquoDelay-dependent observer-based119867
infinfinite-time control for switched systems with time-varying
delayrdquoNonlinear Analysis Hybrid Systems vol 6 no 3 pp 885ndash898 2012
[47] Y Zhang C Liu and X Mu ldquoRobust finite-time119867infincontrol of
singular stochastic systems via static output feedbackrdquo AppliedMathematics and Computation vol 218 no 9 pp 5629ndash56402012
Journal of Applied Mathematics 13
[48] J Cheng H Zhu S Zhong Y Zhang and Y Zeng ldquoFinite-timeStabilization of 119867
infinfiltering for switched stochastic systemsrdquo
Circuits Systems and Signal Processing 2012[49] J Cheng H Zhu S Zhong and Y Zhang ldquoFinite-time bound-
ness of 119867infin
filtering for switching discrete-time systemsrdquo In-ternational Journal of Control Automation and Systems vol 10pp 1129ndash1135 2012
[50] Z Xiang C Qiao and M S Mahmoud ldquoFinite-time analysisand119867
infincontrol for switched stochastic systemsrdquo Journal of the
Franklin Institute vol 349 no 3 pp 915ndash927 2012
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Journal of Applied Mathematics 11
Then we choose 119877 = 119868 119879 = 2 1198881= 1 and 119889 = 001
throughTheorem 15 it yields that themode-dependent filtersare as follows
11986011989111
= [
minus61254 15565
minus02347 minus03763]
11986011989112
= [
minus62682 14338
minus03552 minus05327]
11986011989121
= [
minus89214 24223
minus15476 minus06234]
11986011989122
= [
minus84638 28237
minus13545 minus04322]
11986111989111
= [
51465 minus28516
minus91216 95171]
11986111989112
= [
54203 minus23121
minus93156 96332]
11986111989121
= [
46193 minus11984
minus162397 163006]
11986111989122
= [
46512 minus10311
minus161132 162312]
11986211989111
= [00294 minus00397]
11986211989112
= [00271 minus003368]
11986211989121
= [01163 minus01432]
11986211989122
= [01342 minus03672]
(71)
This paper deals with the finite-time filter design problemfor a class of Markovian jump systems particularly two dif-ferentMarkov processes are considered formodeling the ran-domness of system matrix and the state delay Then throughthe numerical example we can see that results in this paperare feasible which further verified the correctness of our the-ory Therefore the paper shorten this gap
6 Conclusions
In this paper we have examined the problems of finite-time119867
infinfiltering for a class of Markovian jump systems with dif-
ferent system modes Based on a novel approach a sufficientcondition is derived such that the closed-loop Markovianjump system is finite-time bounded and satisfies a prescribedlevel of119867
infindisturbance attenuation in a finite time interval
Finally a numerical example is also given to illustrate theeffectiveness of the proposed design approach It should benoted that one of future research topics would be to investi-gate the problems of fault detection and fault tolerant controlfor time-varying Markovian jump systems with incompleteinformation over a finite-time horizon
Acknowledgments
The authors would like to thank the associate editor and theanonymous reviewers for their detailed comments and sug-gestions This work was supported by the Fund of SichuanProvincial Key Laboratory of Signal and Information Pro-cessing Xihua University (SZJJ2009-002 and SGXZD0101-10-1) and National Basic Research Program of China(2010CB732501)
References
[1] N N Krasovskiı and E A Lidskiı ldquoAnalytical design ofcontrollers in systems with random attributesmdashI Statementof the problem method of solvingrdquo Automation and RemoteControl vol 22 pp 1021ndash1025 1961
[2] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[3] L Wu X Su and P Shi ldquoSliding mode control with bounded1198712gain performance of Markovian jump singular time-delay
systemsrdquo Automatica vol 48 no 8 pp 1929ndash1933 2012[4] H Gao Z Fei J Lam and B Du ldquoFurther results on expo-
nential estimates of Markovian jump systems with mode-de-pendent time-varying delaysrdquo IEEE Transactions on AutomaticControl vol 56 no 1 pp 223ndash229 2011
[5] Z Wu H Su and J Chu ldquo119867infinfiltering for singular Markovian
jump systems with time delayrdquo International Journal of Robustand Nonlinear Control vol 20 no 8 pp 939ndash957 2010
[6] P Tino M Cernansky and L Benuskova ldquoMarkovian architec-tural bias of recurrent neural networksrdquo IEEE Transactions onNeural Networks vol 15 no 1 pp 6ndash15 2004
[7] H Dong Z Wang D W C Ho and H Gao ldquoRobust 119867infin
filtering for Markovian jump systems with randomly occurringnonlinearities and sensor saturation the finite-horizon caserdquoIEEE Transactions on Signal Processing vol 59 no 7 pp 3048ndash3057 2011
[8] Q Ma S Xu and Y Zou ldquoStability and synchronization forMarkovian jump neural networks with partly unknown tran-sition probabilitiesrdquo Neurocomputing vol 74 pp 3404ndash33412011
[9] H Zhao S Xu and Y Zou ldquoRobust119867infinfiltering for uncertain
Markovian jump systems with mode-dependent distributeddelaysrdquo International Journal of Adaptive Control and SignalProcessing vol 24 no 1 pp 83ndash94 2010
[10] Y Zhang S Xu and B Zhang ldquoDiscrete-time fuzzy Markovianjump systems with time-varying delaysrdquo IEEE Transactions onFuzzy Systems vol 17 no 2 pp 411ndash420 2009
[11] Z Wu H Su and J Chu ldquoState estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 no 10ndash12 pp 2247ndash2254 2010
[12] Q Zhu and J Cao ldquoRobust exponential stability of markovianjump impulsive stochastic Cohen-Grossberg neural networkswithmixed time delaysrdquo IEEE Transactions onNeural Networksvol 21 no 8 pp 1314ndash1325 2010
[13] H Huang G Feng and X Chen ldquoStability and stabilizationof Markovian jump systems with time delay via new Lyapunovfunctionalsrdquo IEEE Transactions on Circuits and Systems I vol59 no 10 pp 2413ndash2421 2012
12 Journal of Applied Mathematics
[14] Q Zhu and J Cao ldquoStability of Markovian jump neural net-works with impulse control and time varying delaysrdquoNonlinearAnalysis Real World Applications vol 13 no 5 pp 2259ndash22702012
[15] M Sun J Lam S Xu andY Zou ldquoRobust exponential stabiliza-tion for Markovian jump systems with mode-dependent inputdelayrdquo Automatica vol 43 no 10 pp 1799ndash1807 2007
[16] M Mohammadian A Abolmasoumi and H Momeni ldquo119867infin
mode-independent filter design for Markovian jump geneticregulatory networks with time-varying delaysrdquo Neurocomput-ing vol 87 pp 10ndash18 2012
[17] S Boyd L El Ghaoui E Feron and V Balakrishnan LinearMatrix Inequalities in SystemandControlTheory vol 15 of SIAMStudies in Applied Mathematics SIAM Philadelphia Pa USA1994
[18] S Mou H GaoW Qiang and K Chen ldquoNew delay-dependentexponential stability for neural networks with time delayrdquo IEEETransactions on Systems Man and Cybernetics Part B vol 38no 2 pp 571ndash576 2008
[19] O M Kwon J Park S Lee and E Cha ldquoAnalysis on delay-dependent stability for neural networks with time-varyingdelaysrdquo Neurocomputing vol 103 pp 114ndash120 2013
[20] L Hu H Gao andW X Zheng ldquoNovel stability of cellular neu-ral networkswith interval time-varying delayrdquoNeuralNetworksvol 21 no 10 pp 1458ndash1463 2008
[21] H Zhang Z Liu G B Huang and ZWang ldquoNovel weighting-delay-based stability criteria for recurrent neural networks withtime-varying delayrdquo IEEE Transactions on Neural Networks vol21 no 1 pp 91ndash106 2010
[22] Z Zuo C Yang and YWang ldquoA newmethod for stability anal-ysis of recurrent neural networks with interval time-varyingdelayrdquo IEEE Transactions on Neural Networks vol 21 no 2 pp339ndash344 2010
[23] P Park JWKo andC Jeong ldquoReciprocally convex approach tostability of systems with time-varying delaysrdquo Automatica vol47 no 1 pp 235ndash238 2011
[24] D Zhang L Yu and W Zhang ldquoDelay-dependent fault detec-tion for switched linear systems with time-varying delays-theaverage dwell time approachrdquo Signal Processing vol 91 pp 832ndash840 2011
[25] Z Fei H Gao and P Shi ldquoNew results on stabilization ofMark-ovian jump systems with time delayrdquoAutomatica vol 45 no 10pp 2300ndash2306 2009
[26] H R Karimi ldquoRobust delay-dependent 119867infin
control of uncer-tain time-delay systems with mixed neutral discrete anddistributed time-delays and Markovian switching parametersrdquoIEEE Transactions on Circuits and Systems I vol 58 no 8 pp1910ndash1923 2011
[27] M Luo G Liu and S Zhong ldquoRobust fault detection ofMarkovian jump systems with different systemmodesrdquoAppliedMathematical Modelling vol 37 no 7 pp 5001ndash5012 2013
[28] B Zhang andY Li ldquoExponential1198712minus119871
infinfiltering for distributed
delay systems with Markovian jumping parametersrdquo SignalProcessing vol 93 no 1 pp 206ndash216 2013
[29] A Abolmasoumi and H Momeni ldquoRobust observer-based119867infin
control of a Markovian jump system with different delay andsystem modesrdquo International Journal of Control Automationand Systems vol 9 no 4 pp 768ndash776 2011
[30] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[31] S Ma and E K Boukas ldquoRobust 119867infin
filtering for uncertaindiscrete Markov jump singular systems with mode-dependenttime delayrdquo IET Control Theory amp Applications vol 3 no 3 pp351ndash361 2009
[32] Y ChenW Bi andW Li ldquoStability analysis for neural networkswith time-varying delay a more general delay decompositionapproachrdquo Neurocomputing vol 73 no 4ndash6 pp 853ndash857 2010
[33] Z Shu J Lam and S Xu ldquoRobust stabilization of Markoviandelay systems with delay-dependent exponential estimatesrdquoAutomatica vol 42 no 11 pp 2001ndash2008 2006
[34] X Huang W Lin and B Yang ldquoGlobal finite-time stabilizationof a class of uncertain nonlinear systemsrdquo Automatica vol 41no 5 pp 881ndash888 2005
[35] C Qian and J Li ldquoGlobal finite-time stabilization by outputfeedback for planar systems without observable linearizationrdquoIEEE Transactions on Automatic Control vol 50 no 6 pp 885ndash890 2005
[36] F Amato R Ambrosino C Cosentino and G De TommasildquoInput-output finite time stabilization of linear systemsrdquo Auto-matica vol 46 no 9 pp 1558ndash1562 2010
[37] Y Hong ldquoFinite-time stabilization and stabilizability of a classof controllable systemsrdquo Systems amp Control Letters vol 46 no4 pp 231ndash236 2002
[38] F Amato and M Ariola ldquoFinite-time control of discrete-timelinear systemsrdquo IEEETransactions onAutomatic Control vol 50no 5 pp 724ndash729 2005
[39] Z Zuo Y Liu Y Wang and H Li ldquoFinite-time stochasticstability and stabilisation of linear discrete-time Markovianjump systems with partly unknown transition probabilitiesrdquoIET Control Theory amp Applications vol 6 no 10 pp 1522ndash15262012
[40] X Luan F Liu and P Shi ldquoFinite-time filtering for non-linearstochastic systems with partially known transition jump ratesrdquoIET Control Theory amp Applications vol 4 no 5 pp 735ndash7452010
[41] F Amato M Ariola and C Cosentino ldquoFinite-time control ofdiscrete-time linear systems analysis and design conditionsrdquoAutomatica vol 46 no 5 pp 919ndash924 2010
[42] X Lin H Du and S Li ldquoFinite-time boundedness and 1198712-
gain analysis for switched delay systems with norm-boundeddisturbancerdquo Applied Mathematics and Computation vol 217no 12 pp 5982ndash5993 2011
[43] J Lin S Fei and Z Gao ldquoStabilization of discrete-time switchedsingular time-delay systems under asynchronous switchingrdquoJournal of the Franklin Institute vol 349 no 5 pp 1808ndash18272012
[44] H Song L Yu D Zhang and W-A Zhang ldquoFinite-time 119867infin
control for a class of discrete-time switched time-delay systemswith quantized feedbackrdquoCommunications in Nonlinear Scienceand Numerical Simulation vol 17 no 12 pp 4802ndash4814 2012
[45] Z Zuo H Li Y Liu and Y Wang ldquoOn finite-time stochasticstability and stabilization ofMarkovian jump systems subject topartial information on transition probabilitiesrdquoCircuits Systemsand Signal Processing 2012
[46] H Liu Y Shen and X Zhao ldquoDelay-dependent observer-based119867
infinfinite-time control for switched systems with time-varying
delayrdquoNonlinear Analysis Hybrid Systems vol 6 no 3 pp 885ndash898 2012
[47] Y Zhang C Liu and X Mu ldquoRobust finite-time119867infincontrol of
singular stochastic systems via static output feedbackrdquo AppliedMathematics and Computation vol 218 no 9 pp 5629ndash56402012
Journal of Applied Mathematics 13
[48] J Cheng H Zhu S Zhong Y Zhang and Y Zeng ldquoFinite-timeStabilization of 119867
infinfiltering for switched stochastic systemsrdquo
Circuits Systems and Signal Processing 2012[49] J Cheng H Zhu S Zhong and Y Zhang ldquoFinite-time bound-
ness of 119867infin
filtering for switching discrete-time systemsrdquo In-ternational Journal of Control Automation and Systems vol 10pp 1129ndash1135 2012
[50] Z Xiang C Qiao and M S Mahmoud ldquoFinite-time analysisand119867
infincontrol for switched stochastic systemsrdquo Journal of the
Franklin Institute vol 349 no 3 pp 915ndash927 2012
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
12 Journal of Applied Mathematics
[14] Q Zhu and J Cao ldquoStability of Markovian jump neural net-works with impulse control and time varying delaysrdquoNonlinearAnalysis Real World Applications vol 13 no 5 pp 2259ndash22702012
[15] M Sun J Lam S Xu andY Zou ldquoRobust exponential stabiliza-tion for Markovian jump systems with mode-dependent inputdelayrdquo Automatica vol 43 no 10 pp 1799ndash1807 2007
[16] M Mohammadian A Abolmasoumi and H Momeni ldquo119867infin
mode-independent filter design for Markovian jump geneticregulatory networks with time-varying delaysrdquo Neurocomput-ing vol 87 pp 10ndash18 2012
[17] S Boyd L El Ghaoui E Feron and V Balakrishnan LinearMatrix Inequalities in SystemandControlTheory vol 15 of SIAMStudies in Applied Mathematics SIAM Philadelphia Pa USA1994
[18] S Mou H GaoW Qiang and K Chen ldquoNew delay-dependentexponential stability for neural networks with time delayrdquo IEEETransactions on Systems Man and Cybernetics Part B vol 38no 2 pp 571ndash576 2008
[19] O M Kwon J Park S Lee and E Cha ldquoAnalysis on delay-dependent stability for neural networks with time-varyingdelaysrdquo Neurocomputing vol 103 pp 114ndash120 2013
[20] L Hu H Gao andW X Zheng ldquoNovel stability of cellular neu-ral networkswith interval time-varying delayrdquoNeuralNetworksvol 21 no 10 pp 1458ndash1463 2008
[21] H Zhang Z Liu G B Huang and ZWang ldquoNovel weighting-delay-based stability criteria for recurrent neural networks withtime-varying delayrdquo IEEE Transactions on Neural Networks vol21 no 1 pp 91ndash106 2010
[22] Z Zuo C Yang and YWang ldquoA newmethod for stability anal-ysis of recurrent neural networks with interval time-varyingdelayrdquo IEEE Transactions on Neural Networks vol 21 no 2 pp339ndash344 2010
[23] P Park JWKo andC Jeong ldquoReciprocally convex approach tostability of systems with time-varying delaysrdquo Automatica vol47 no 1 pp 235ndash238 2011
[24] D Zhang L Yu and W Zhang ldquoDelay-dependent fault detec-tion for switched linear systems with time-varying delays-theaverage dwell time approachrdquo Signal Processing vol 91 pp 832ndash840 2011
[25] Z Fei H Gao and P Shi ldquoNew results on stabilization ofMark-ovian jump systems with time delayrdquoAutomatica vol 45 no 10pp 2300ndash2306 2009
[26] H R Karimi ldquoRobust delay-dependent 119867infin
control of uncer-tain time-delay systems with mixed neutral discrete anddistributed time-delays and Markovian switching parametersrdquoIEEE Transactions on Circuits and Systems I vol 58 no 8 pp1910ndash1923 2011
[27] M Luo G Liu and S Zhong ldquoRobust fault detection ofMarkovian jump systems with different systemmodesrdquoAppliedMathematical Modelling vol 37 no 7 pp 5001ndash5012 2013
[28] B Zhang andY Li ldquoExponential1198712minus119871
infinfiltering for distributed
delay systems with Markovian jumping parametersrdquo SignalProcessing vol 93 no 1 pp 206ndash216 2013
[29] A Abolmasoumi and H Momeni ldquoRobust observer-based119867infin
control of a Markovian jump system with different delay andsystem modesrdquo International Journal of Control Automationand Systems vol 9 no 4 pp 768ndash776 2011
[30] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[31] S Ma and E K Boukas ldquoRobust 119867infin
filtering for uncertaindiscrete Markov jump singular systems with mode-dependenttime delayrdquo IET Control Theory amp Applications vol 3 no 3 pp351ndash361 2009
[32] Y ChenW Bi andW Li ldquoStability analysis for neural networkswith time-varying delay a more general delay decompositionapproachrdquo Neurocomputing vol 73 no 4ndash6 pp 853ndash857 2010
[33] Z Shu J Lam and S Xu ldquoRobust stabilization of Markoviandelay systems with delay-dependent exponential estimatesrdquoAutomatica vol 42 no 11 pp 2001ndash2008 2006
[34] X Huang W Lin and B Yang ldquoGlobal finite-time stabilizationof a class of uncertain nonlinear systemsrdquo Automatica vol 41no 5 pp 881ndash888 2005
[35] C Qian and J Li ldquoGlobal finite-time stabilization by outputfeedback for planar systems without observable linearizationrdquoIEEE Transactions on Automatic Control vol 50 no 6 pp 885ndash890 2005
[36] F Amato R Ambrosino C Cosentino and G De TommasildquoInput-output finite time stabilization of linear systemsrdquo Auto-matica vol 46 no 9 pp 1558ndash1562 2010
[37] Y Hong ldquoFinite-time stabilization and stabilizability of a classof controllable systemsrdquo Systems amp Control Letters vol 46 no4 pp 231ndash236 2002
[38] F Amato and M Ariola ldquoFinite-time control of discrete-timelinear systemsrdquo IEEETransactions onAutomatic Control vol 50no 5 pp 724ndash729 2005
[39] Z Zuo Y Liu Y Wang and H Li ldquoFinite-time stochasticstability and stabilisation of linear discrete-time Markovianjump systems with partly unknown transition probabilitiesrdquoIET Control Theory amp Applications vol 6 no 10 pp 1522ndash15262012
[40] X Luan F Liu and P Shi ldquoFinite-time filtering for non-linearstochastic systems with partially known transition jump ratesrdquoIET Control Theory amp Applications vol 4 no 5 pp 735ndash7452010
[41] F Amato M Ariola and C Cosentino ldquoFinite-time control ofdiscrete-time linear systems analysis and design conditionsrdquoAutomatica vol 46 no 5 pp 919ndash924 2010
[42] X Lin H Du and S Li ldquoFinite-time boundedness and 1198712-
gain analysis for switched delay systems with norm-boundeddisturbancerdquo Applied Mathematics and Computation vol 217no 12 pp 5982ndash5993 2011
[43] J Lin S Fei and Z Gao ldquoStabilization of discrete-time switchedsingular time-delay systems under asynchronous switchingrdquoJournal of the Franklin Institute vol 349 no 5 pp 1808ndash18272012
[44] H Song L Yu D Zhang and W-A Zhang ldquoFinite-time 119867infin
control for a class of discrete-time switched time-delay systemswith quantized feedbackrdquoCommunications in Nonlinear Scienceand Numerical Simulation vol 17 no 12 pp 4802ndash4814 2012
[45] Z Zuo H Li Y Liu and Y Wang ldquoOn finite-time stochasticstability and stabilization ofMarkovian jump systems subject topartial information on transition probabilitiesrdquoCircuits Systemsand Signal Processing 2012
[46] H Liu Y Shen and X Zhao ldquoDelay-dependent observer-based119867
infinfinite-time control for switched systems with time-varying
delayrdquoNonlinear Analysis Hybrid Systems vol 6 no 3 pp 885ndash898 2012
[47] Y Zhang C Liu and X Mu ldquoRobust finite-time119867infincontrol of
singular stochastic systems via static output feedbackrdquo AppliedMathematics and Computation vol 218 no 9 pp 5629ndash56402012
Journal of Applied Mathematics 13
[48] J Cheng H Zhu S Zhong Y Zhang and Y Zeng ldquoFinite-timeStabilization of 119867
infinfiltering for switched stochastic systemsrdquo
Circuits Systems and Signal Processing 2012[49] J Cheng H Zhu S Zhong and Y Zhang ldquoFinite-time bound-
ness of 119867infin
filtering for switching discrete-time systemsrdquo In-ternational Journal of Control Automation and Systems vol 10pp 1129ndash1135 2012
[50] Z Xiang C Qiao and M S Mahmoud ldquoFinite-time analysisand119867
infincontrol for switched stochastic systemsrdquo Journal of the
Franklin Institute vol 349 no 3 pp 915ndash927 2012
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Journal of Applied Mathematics 13
[48] J Cheng H Zhu S Zhong Y Zhang and Y Zeng ldquoFinite-timeStabilization of 119867
infinfiltering for switched stochastic systemsrdquo
Circuits Systems and Signal Processing 2012[49] J Cheng H Zhu S Zhong and Y Zhang ldquoFinite-time bound-
ness of 119867infin
filtering for switching discrete-time systemsrdquo In-ternational Journal of Control Automation and Systems vol 10pp 1129ndash1135 2012
[50] Z Xiang C Qiao and M S Mahmoud ldquoFinite-time analysisand119867
infincontrol for switched stochastic systemsrdquo Journal of the
Franklin Institute vol 349 no 3 pp 915ndash927 2012
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of