Download - MR 댐퍼를 이용한 지진하중을 받는 지진격리 벤치마크 구조물의 신경망제어
MR 댐퍼를 이용한 지진하중을 받는지진격리 벤치마크 구조물의 신경망제어
MR 댐퍼를 이용한 지진하중을 받는지진격리 벤치마크 구조물의 신경망제어
Heon-Jae Lee*: Ph.D. CandidatePh.D. Candidate, KAIST, Korea
Sang-Won Cho: Post Doctoral Fellow, UWO, Canada
Ju-Won Oh: Professor, Hannam University, Korea
In-Won Lee: Professor, KAIST, Korea
2006 2006 춘계 지진공학회 학술발표회춘계 지진공학회 학술발표회17~18, Mar, 200617~18, Mar, 2006
22Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관
• IntroductionIntroduction
• Benchmark ProblemBenchmark Problem
• Proposed MethodProposed Method
• Training Neuro-Control SystemTraining Neuro-Control System
• Performance EvaluationPerformance Evaluation
• ConclusionConclusion
OUTLINEOUTLINE
33Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관Introduction
• Base Isolation– One of the most widely implemented seismic
protection system– Mitigates the effects of an earthquake by isolating
the structure from ground motion
Kodiak, Alaska San Francisco
44Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관
• Nonlinear Devices for Base Isolation
Benefit– Restoring force and adequate damping
capacity can be obtained in one device
Drawbacks– Strongly nonlinear
– Poor adaptability for a wide range of ground motion
– Especially strong impulsive ground motions generated at near-source location
Lead Rubber Bearing
Friction Pendulum Bearing
Introduction
55Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관
• Hybrid Control Strategies– Consisting of a base isolation system combined
with actively controlled actuators
– Advantages• High performance in reducing vibration• High adaptability for different ground excitation• Ability to control of multiple vibration modes
– Drawbacks• Requirement of a large external power supply• Active systems may have the risk of instability
Introduction
66Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관
• Semi-Active Base Isolation System– Consisting of a base isolation system that
employs semi-active control devices (e.g. MR
dampers, controllable friction dampers)
– Similar high adaptability to the active system
– No requirement of large power supplies
Introduction
77Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관Introduction
• Semi-Active Neuro-Controller– Improved neuro-controller (Kim et al., 2000, 2001)
• New training algorithm based on cost function• Sensitivity evaluation algorithm to replace an emulator neural
network
– Clipped algorithm• Clips the control force that cannot be achieved by MR dampe
r
88Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관
• Purpose– To provide systematic and standardized means – By making direct comparisons between
competing control strategies, including devices, algorithms, sensors, etc.
– To allow researchers in structural control to test their algorithms and devices
– This benchmark problem is about a smart base isolation system.
Benchmark Problem
99Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관
• Benchmark Structure– A base-isolated eight-story,
steel-braced framed building– Similar to existing building in
Los Angeles, California
– In this investigation, only the linear elastomeric isolation system which consists of 92 bearings is considered.
– Modeled using three master degrees of freedom per floor
– 16 MR dampers (8 along the x-axis and 8 along the y-axis) are also installed
Benchmark Problem
1010Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관
• Sample Earthquakes– Both the fault-normal (FN) and fault-parallel (FP) c
omponents of Newhall, Sylmar, El Centro, Rinaldi, Kobe, Ji-ji, Erzinkan
• Control Cases– Case I : x-direction (FP), y-direction (FN)– Case II: x-direction (FN), y-direction (FP)
Benchmark Problem
1111Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관
STRUCTURE
Neural Network
xx ,gx
fMR Damper
Clipped
Algorithmdf
v
Clipped Neuro-Controller
Block diagram of the proposed method
Proposed Method
1212Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관
1I
2I
1nI 23no
22o
21o
1ihW
2jiW
• Structure of the neural network• Structure of the neural network
Inputlayer
Hiddenlayer
Outputlayer
Proposed Method
1414Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관
• Improved Neuro-Controller– New training algorithm
• The neuro-controller is trained by minimizing a cost function
1
0k
T
k1k
T
1 RuuyQy2
1ˆN
kkJ
where, is specific state, is control signal and
are weighting matrices
y u
R,Q
Proposed Method
1616Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관
• Improved Neuro-Controller– The update of weights and biases between the output lay
er and hidden layer can be simply expressed as
T
iji oW 122 22 jb
where,
21T
1
2 1Ruu
yQy fT
k
k
kk
:learning rate
:sensitivity
:unit vector
:activation function of the output layer
kk uy 1
12f
Proposed Method
1717Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관
• Improved Neuro-Controller– In the same manner, update of those between the hid
den layer and input layer can be obtained as
T
hihW I11 11 ib
1221 fWT
ji where,
Proposed Method
1818Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관Proposed Method
• Clipped Algorithm– Desired force (by neural network):– Generated force (by MR damper):
df
f
df
f
0v0v
0v
maxVv
maxVv maxVv
maxVv
0v
1919Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관
• Training Data– Filtered artificial earthquake record– Magnitude: scaled to match the maximum acceleration
of the given earthquakes
– Shaping filter
22 2
4)(
ggg
gg
ss
ssF
where,
3.0,2 gg
0 5 10 15 20 25 30-10
-5
0
5
10
Time (sec)
Gro
und
acce
lera
tion
(m/s
ec2
)
Training Neuro-Controller
2020Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관
• Structure of Neuro-Controller– Two neuro-controllers are employed for x and y-dire
ctions, respectively.
bx
bx
8x
bdx
gx
df
Structure of each direction’s neuro-controller
Training Neuro-Controller
2121Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관Training Neuro-Controller
• Cost Function– In cost function, are included.– Optimal weighting matrices
1510
)001.0,1.0,1.0(
r
diagQ
bb xxx ,,8
2222Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관
• Cost function vs. epoch
– The cost function converges in both x and y-directions, which means the training is successful.
2 4 6 8 10 12 14 16 18 200
500
1000
1500
2000
2500
3000
3500
epoch
Cos
t fun
ctio
n
x-direction
y-direction
Training Neuro-Controller
2323Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관
• Evaluation Criteria– Based on both maximum and RMS responses of the
building
J1 - Peak base shear
J2 - Peak structure shear
J3 - Peak base displacement
J4 - Peak inter-story drift
J5 - Peak absolute floor acceleration
J6 - Peak generated force
J7 - RMS base displacement
J8 - RMS absolute floor acceleration
J9 - Total energy absorbed
),(ˆmax
),(max
0
01
qtV
qtVJ
),(ˆmax
),(max
1
12
qtV
qtVJ
),(ˆmax
),(max3
qtd
qtdJ
i
i),(ˆmax
),(max4
qtd
qtdJ
f
f
),(ˆmax
),(max5
qta
qtaJ
f
f
),(max
),(max
06 qtV
qtFJ
k
),(max
),(max
ˆ
7qt
qtJ
d
d
),(max
),(max
ˆ8
qt
qtJ
a
a
dtqtUqtV
dtqtvqtFJ
g
kk
),(),(
),(),(
09
Performance Evaluation
2424Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관Performance Evaluation
• Performance Comparison (Case I)
00.20.40.60.81
1.21.41.6
1 2 3 4 5 6 7 8 9
Sylmar
0
0.5
1
1.5
2
2.5
3
1 2 3 4 5 6 7 8 9
El Centro
00.20.40.60.81
1.21.41.6
1 2 3 4 5 6 7 8 9
Rinaldi
0
0.2
0.4
0.6
0.8
1
1 2 3 4 5 6 7 8 9
Jiji
0
0.2
0.4
0.60.8
1
1.2
1.4
1 2 3 4 5 6 7 8 9
Erzinkan
0
0.5
1
1.5
2
2.5
1 2 3 4 5 6 7 8 9
Kobe
0
0.5
1
1.5
2
2.5
3
1 2 3 4 5 6 7 8 9
Newhall
Passive (Vmax)
Clipped optimal
Proposed
2525Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관Performance Evaluation
0
0.5
1
1.5
2
1 2 3 4 5 6 7 8 9
0
0.20.4
0.60.8
11.2
1.4
1 2 3 4 5 6 7 8 90
0.51
1.52
2.53
3.54
1 2 3 4 5 6 7 8 9
0
0.2
0.4
0.6
0.8
1
1.2
1 2 3 4 5 6 7 8 90
0.5
1
1.5
2
2.5
1 2 3 4 5 6 7 8 90
0.2
0.4
0.6
0.8
1
1 2 3 4 5 6 7 8 9
0
0.2
0.4
0.60.8
1
1.2
1.4
1 2 3 4 5 6 7 8 9
Sylmar El Centro
Rinaldi
Jiji
Erzinkan
Kobe
Newhall
• Performance Comparison (Case II)
Passive (Vmax)
Clipped optimal
Proposed
2626Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관
• Corner drift (normalized by uncontrolled values)
Case I Case II
Passive Clipped Proposed Passive Clipped Proposed
Newhall 1.50 1.28 1.01 1.21 1.11 0.74
Sylmar 0.76 0.96 0.75 0.86 0.90 0.71
El Centro 1.10 0.93 0.57 2.06 1.30 0.91
Rinaldi 0.91 0.89 0.84 0.89 0.93 0.94
Kobe 1.22 1.07 0.80 1.06 1.10 0.85
Jiji 0.80 0.78 0.77 0.96 0.92 1.01
Erzinkan 0.72 0.73 0.65 0.90 0.65 0.74
Performance Evaluation
2727Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관
• A new semi-active control strategy for seismic response reduction using neuro-controller and MR dampers is proposed.
• The proposed strategy was applied to a benchmark building installed with linear elastomeric isolation system.
• In numerical simulation results, the proposed strategy can significantly reduce the floor acceleration, base shear and building corner drift with only a slight increase of the base displacement.
Conclusion
2828Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관
Thank you for your attention!!!
2929Structural Dynamics & Vibration Control Lab., KAIST, KoreaStructural Dynamics & Vibration Control Lab., KAIST, Korea
2006 한국지진공학회 학술발표회한국과학기술원 창의학습관
• Sensitivity evaluation algorithm (Kim et al., 2001)• Sensitivity evaluation algorithm (Kim et al., 2001)
BuAzz (9)
In the discrete-time domain,
kkk HuGzz 1
sTs eT A)G(
BA)I()H( -1A sTs eT
(10)
where
represents the sensitivity, .H
jk
k
,
1
u
z
kkk HuGzz 1
State space equation of structure