real time control logical architecture discussion of logical architecture of real time control...
TRANSCRIPT
Real time controlLogical architecture
Discussion of logical architecture of real time control systems
M.Jonker
Objectives
• Discuss real time control systems from a different perspective. I.e do not consider– why we need a real time system
– how we can implement the real time system
• A discussion of the logical architecture will highlight some of the basic ingredients.
• At the end we should be able to see the problem from a larger perspective.
Objectives
• We will be able to find general concepts that can be specialized into solutions for specific problems. (Generalizing specific solutions may not always work).
• We will find new or different criteria for the physical implementation of a real time control system.
Data Processing & Summing
Control knob processingControl knob
processing
Data Processing & Summing
Control knob processing
Objectives: be able to understand this
Sources
Measurement Data reduction
Settings database
SourcesSourcesSourcesSources
Other Intermediate Data Processing
Data Surveillance
Reference generators
ControlControlControlControl
Feedback processing
RT knob inputs
Data Display
Reference generators
Reference generators
Feedback processing
Feedback processing
Measurement Data reduction
Measurement Data reduction
Beam Abort
Reference Function & settingsTiming
Feed-ForwardFeed-Forward
Objectives: be able to understand this
IQf IQd
MQMQ
Mij(E)function of energy given by the control system
E(t)Energy function of ramp time given by the control system
Start
Stop
tramp
ramp time generator
Qh
Qv
Filter
Filter
IQd(t)
IQd reference
function given by the control system
IQf(t)
IQf reference
function given by the control system
Qv(t)tune reference given by the control system
Qh(t)tune reference given by the control system
Qh
Real Time Knob
Do we need a real time control system...
Can we avoid it?• If we deny the need for a real time control
system we will end up with several real time systems ‘provided’ as specific solutions.– BI real time acquisition systems (BST, ATM?– PO (GPS-time Gbit Ethernet)– Multipole factory– RF ...
This will not necessarily provide an adequate framework for implementing a coherent control system.
Who is afraid of thereal time control ?
Real time does not mean fast and expensive.
It implies reliable and deterministic.There may be many reliable and affordable solutions possible. But the implementation should not be driven by the solutions.
Control: measurement, interpretation, correction– What are the measurements– Who is interpreting the data– How do we make the correction
Real-time: deterministic– what are the required response times of measurement, interpretation and
correction.
disclaimer• I personally find the real time control system a very
fascinating subject.• But, I do not have the time (yet) to get deeply
involved with this.• However, I think it is important to share my
perspectives with you because, it will help to advance in the ever ongoing real-time controls dilemma.
• finally, I want to make sure that this gets on the right track, so in a few years it will be a pleasure to join you in the real time control.
The following slides are taken from two previous un-presentations:
LHC-RT-control-brainstorming.ppt In preparation of the LHC RT controls workshop 13-April 2000
RT-Architecture-brainstorming.ppt25-October 2000
RT-Control ArchitectureIf we need to implement an RT control somewhere, we should not
only concentrate on specific requirement. It will be very helpful to see the problem in a larger perspective, i.e. as a specialisation of a more general problem.
The aim of this presentation is to:• concentrate on the essential elements taking part in the control structure.• identify common control components.• provide a framework for further discussions on performance and
implementation issues.
In this context we should not address realisation and implementation issues like:
• where we need the RT-control• what the performance should be of the RT-control
Control connectivity
Please note:
The next slides show examples of connectivity in increasing complexity. It gives a logical view of how different components participating in the control are related.
This should not be taken as a proposal for an actual implementation.
• There is no assumption made about the actual location of the various components.
• Certain control components with the same function are only drawn once (e.g. ramp time generator). This does not exclude that certain of this functional components may be duplicated in the actual implementation of the control system.
Control connectivity
Observation
Control
Observation
Control
Transformation
DataOut
DataIn
DataIn-DataOut
Connections:• Data may be more than one parameter• DataOut may be connected to more than one data-in• DataIn may be connected to (i.e. get its data from) more than
one DataOut (we assume the data will be summed)
Mij
constants given by the control system
X MQ
Control connectivity
Mij(t)function of ramp time given by the control system
Start
StopMQMQ
Mij(E)function of energy given by the control system
E(t)Energy function of ramp time given by the control system
Start
Stop
tramp
ramp time generator
Filter
Filter
IQd(t)
IQd reference
function given by the control system
IQf(t)
IQf reference
function given by the control system
Qv(t)tune reference given by the control system
Qh(t)tune reference given by the control system
Qh
Real Time Knob
IQdIQf
Qh
Qv
Real time controlConnectivity of feedback loops:
Measurem
entequipm
ent
Control EquipmentBeam
Real time control
Measurem
entequipm
ent
Control EquipmentBeam
By proper modelling, these blobs can often be controlled by a single or a few parameters
• Connectivity of feedback loops:
Real time control
Connectivity of feedback loops:
Measurem
entequipm
ent
Control EquipmentBeam
Control connectivity
PU1
Corr.1 Corr.2
PU2 PU3 PU4 PUn
Calculate OrbitBump
Corr.3
Filter
Start
Stop
tramp
ramp time generator
Bump(t)Bump reference function
E(t)Energy function of ramp time
Corr.1(t)reference functionCorr.1(t)reference function
Corr.3(t)reference function
Calculate OrbitBump corrections
Control connectivity
Is function generation inside or outside the digital controller ?This is part of the physical architecture
sum
MQ Start
Stop
tramp
ramp time generatorIQf(t)
IQf reference
function given by the control system
IQf
Control connectivity ingredients
Data Destinations:•Data Transformations•Controllers•RT displays
Data Transformation, Data in - Data out:•Reference functions•Multipole factories•Filters (Feedback)•Tune Matrices•Orbit calculation
Data transformations will need setting parameters from the controls system
Data sources:•Measurement•RT-Knobs•Data Transformations
RT control can be considered as a network of:
RT- connectivity:•RT backbone•RT configuration manager
RT control issues
Hence we should answer these questions:
• Which Data sources– How much data, what frequency, potentially useful for doing what, and who said so.
• Which Control targets– How much data, what frequency, potentially useful for doing what, and who said so.
• Which Data Destinations, Control Origins, Feedback Applications– How much latency, need for synchronisation, etc.
• Do we need a configurable RT system? What are the advantages, are there any disadvantages?
• Is there in-house experience, commercial solutions, RT middleware ?
Identify real-time needs
data sources, data destinations, control objects, control data sources, processing (feedback loops)
Sources Control
Destinations Origins
Loops
Settings database
RT knobs,timing
Reference values
Operator
Data Processing & Summing
Control knob processingControl knob
processing
Data Processing & Summing
Control knob processing
Identification of real-time needs
Sources
Measurement Data reduction
Settings database
SourcesSourcesSourcesSources
Other Intermediate Data Processing
Data Surveillance
Reference generators
ControlControlControlControl
Feedback processing
RT knob inputs
Data Display
Reference generators
Reference generators
Feedback processing
Feedback processing
Measurement Data reduction
Measurement Data reduction
Beam Abort
Reference Function & settingsTiming
Feed-ForwardFeed-Forward
Identify real-time needs
• real-time data sources– Local orbit– Radial Position– Global Orbit– Beam Size– Beam Current– Beam Loss– Tune– Chromaticity– Beam Beam deflection– Luminosity
You name it, we will control something with it?
– Synchronisation
Identify real-time needs
• real-time data destinations– on-line displays
– Beam Dump?
Identify real-time needs
• real-time control objects– Power converters
– RF equipment
– Reference settings for local feedback loops
– Reference settings for RT surveillance
– Measurement parameters
Could we have one type of device model (or even better: hardware) for control please?
Identify real-time needs
• real-time control data sources– ramp functions and actual settings changes (trims)
– feed forward data
– feed back loops
– RT knobs
• Where do they get the data from– RT input (timing, measurement, operator tools)
– Controls data base with trim archival etc
Identify real-time needs• real time processing
– Operator: universal control, but real-time?– Application:
• feedback loops
• data surveillance
– Note: RT processing could be distributed and partly included in the measurement, and/or in the control equipment. Example:
• A local orbit bump is measured with 5 pick-ups, the measurement system transforms these 5 signals into a single bump amplitude.
• A bump correction ‘knob’ is defined that corrects the bump amplitude using three correctors.
• A feedback loop takes the measured bump amplitude and regulates the feedback by controlling the bump knob. The feedback process has the responsibility to provide a stable regulation in time domain.
Logical Architecture Recommendations (to be reviewed and studied).
Allow an efficient and flexible way to connect and parameterise the objects involved in RT control:– measurement, processing, control
– RT network with RT agents. A system of “standard” rt-processes that can be reconfigured with configurable deterministic data channels for communication.
Physical Architecture Recommendations (to be reviewed and studied).
• The RT-architecture should not fix its application for the coming N years.
• The architecture should be open. It should allow implementing what is required today, and what may be asked tomorrow.– Example Generic reference generation agent to replace function
generation capability in the electronics. Can be used for PC control, feedback reference, surveillance reference, etc.
Physical architecture… or the art of building pyramids
PO BI
CO