1 / 43

From Linear Systems to Discrete-Event Systems

From Linear Systems to Discrete-Event Systems. W.M. Wonham Systems Control Group ECE Department University of Toronto Xidian University – Dec 2006 Update 2006.11.22. What is a Discrete-Event System?.

selena
Download Presentation

From Linear Systems to Discrete-Event Systems

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. From Linear Systems to Discrete-Event Systems W.M. Wonham Systems Control Group ECE Department University of Toronto Xidian University – Dec 2006 Update 2006.11.22

  2. What is a Discrete-Event System? • Structure with ‘states’ having duration in time, ‘events’ happening instantaneously and asynchronously. • States: machine is idle, is operating, is broken down, is under repair. • Events: machine starts work, breaks down, completes work or repair. • State space discrete in time and space. • State transitions ‘labeled’ by events.

  3. Summary • Some history • Supervisory Control Theory (SCT) • Large systems (using IDDs) • Hierarchy • Extensions and Applications • Conclusions

  4. Systems Control Concepts (c. 1980) • State space framework well-established: Controllability Observability Optimality (Quadratic, Lvarious, H) • Qualitative synthesis via controlled dynamic invariants • Use of geometric constructs and partial order: Controllability subspaces (c.s.) - supremal subspaces!

  5. Discrete-Event Systems (c. 1980) • Practical problems • Programming languages for modeling & simulation • Queues, Markov chains • Petri nets • Boolean models • Formal languages • Process algebra (CSP, CCS)

  6. Discrete-Event Systems Control (c.1980) • Control problems implicit in the literature (enforcement of resource constraints, synchronization, ...) But • Emphasis on modeling, simulation, verification • Little formalization of control synthesis • Absence of control-theoretic ideas • No standard model or approach to control

  7. Needed (1980): DES Control Theory • System model Discrete in time and (usually) space Asynchronous (event-driven) Nondeterministic - support transitional choices • Amenable to formal control synthesis - exploit control concepts • Applicable: manufacturing, traffic, logistic,...

  8. Proposed (1982):Supervisory Control Theory(Ramadge & Wonham) • Automaton representation - internal state descriptions for concrete modeling and computation • Language representation - external i/o descriptions for implementation-independent concept formulation • Simple control‘technology’

  9. Community Response Anonymous Referees (1983-87) • Automatica “Automata have no place in control engineering.” Reject! • Mathematical Systems Theory • “Finite automata and regular languages are • nothing new at best and trivial at worst.” • Reject! • SIAM J. Control & Optimization • “So this is optimal control? Well...”Accept

  10. Summary • Some history • Supervisory Control Theory (SCT) • Large systems (using IDDs) • Hierarchy • Extensions and Applications • Conclusions

  11. Automaton controllable Idle  MACH   Down Wkg  • Control Technology • = {, }  {, } = con  uncon uncontrollable SCT Base Model

  12. 0 10 13 11 1 2 12 TCTMACH MACH := (Q, , , q0, Qm) MACH = Create (MACH) >name:MACH ># states:3 {TCT Q := {0,1,2}, q0 := 0} >marker state(s): 0 {TCT Qm := {0}} > transitions:[0,11,1], [1,10,0], [1,12,2], [2,13,0] {TCT := {10, 11, 12, 13},  : Q    Q transitions} > quit<Ret> {TCT files MACH.DES}

  13. I    W D  SCT Languages • Closed and Marked Behaviors L(MACH) = all strings generable from initial state I = {, , , , , , …} = closed behavior of MACH Lm(MACH) = all generable strings hitting some marker state = {, , , …} = marked behavior of MACH prefix closure • _________ • Liveness (Nonblocking):Lm(MACH) = L(MACH)

  14. Synchronous Product • Builds a more complex automaton with more complex language          shared L(A1  A2) = P1-1L(A1)  P2-1 L(A2) expressed bynatural projections Pi: (1 2)*  i* (i = 1,2)

  15. Transfer Line TL (Al-Jaar & Desrochers) 1 4 5 6 2 3 M1 B1 B2 TU M2 8 SCT Complex Plant • Complex plant = sync product of simplesubplants TL = M1 || M2 || TU

  16. 2, 8 2, 8 2,8 B1 3 3 3 4 B2 5 SCT Complex (Safety) Specification • Complex specification = sync product of partial specifications BUFFSPEC = B1 || B2

  17. General Control Issues • Is there a control that enforces bothsafety, and liveness (nonblocking),and which is maximally permissive ? • If so, can its design be automated? • If so, with acceptable computingeffort?

  18. SCT Synthesis - Problem E.g. for TL, let ConTL = ‘TL under control’ Must guarantee 1. Safety: Lm(ConTL)  Lm(BUFFSPEC) 2. Liveness (nonblocking):  Lm(ConTL) = L(ConTL) 3. Maximal permissiveness: Lm(ConTL) = maximum subject to safety and liveness

  19. SCT Synthesis - Solution E.g. for TL: 1. Fundamental definition A sublanguage K Lm(TL) is controllable if __ K uncon L(TL)  K  “Once in K, you can’t skid out on an uncontrollable event.” 2. Fundamental result There exists a (unique)supremalcontrollable sublanguage Ksup Lm(TL)  Lm(BUFFSPEC) Furthermore Ksup can be effectively computed.

  20. * (all strings) Lm(TL) Lm(BUFFSPEC) Lm(TL)  Lm(BUFFSPEC) optimization Ksup (optimal) K' K" (suboptimal)  (no strings) SCT Synthesis Lattice

  21. ‘Monolithic’ SCT Implementation • Given TL and BUFFSPEC, compute Ksup Ksup= Lm(SUPER) SUPER = supcon (TL, BUFFSPEC) •Given SUPER, implement Ksup TL Ksup enable/disable events in con SUPER

  22. TCT TRANSFER LINE (TL) M1 = Create (M1), M2 = Create (M2), TU = Create (TU) TL = Sync (M1, M2, TU) {synchronous product} B1 = Create (B1), B2 = Create (B2) BUFFSPEC = Sync (B1, B2) {synchronous product} SUPER (.DES) = SupCon (TL, BUFFSPEC) {optimization} SUPER (.DAT) = ConDat (TL, SUPER(.DES)) {control data} SIMSUP = SupReduce (TL, SUPER(.DES), SUPER(.DAT)) {supervisor reduction} SIMSUP (.DAT) = ConDat (TL, SIMSUP) {control data}

  23. Summary • Some history • Supervisory Control Theory (SCT) • Large systems (using IDDs) • Hierarchy • Extensions and Applications • Conclusions

  24. PLANT = sync (PLANT.1, … , PLANT.m) SPEC = sync (SPEC.1, … , SPEC.n) SUPER = supcon (PLANT, SPEC) State size of SUPER~ (Constant) m+n Exponential state space explosion ! ‘Extensional’ listing of ‘flat’ transition structures is impossible ! Large DES

  25. What To Do ? • In state representations, retain product structurePLANT state vector x = [x1, ... , xm]SPEC state vector y = [y1, … , yn] • Express SUPER as a predicate Predsup (x, y, , x, y) = 0 or 1 • Algorithmize representation of Predsup using Integer Decision Diagrams (IDDs)

  26. Root x1 0 1 2 Order! x2 0 1 0 0 1 1 f 1 0 1 0 0 0 Reduce! Root x1 0 2 1 IDD x2 0 0 1 f 1 0 Integer Decision Diagrams (IDDs) • IDDs represent functions on finite sets

  27. Manufacturing Workcell (Barkaoui & Ben Abdallah 1995, Seidl 2000) Input 1 Output 2 Machine 2 Machine 1 Robot 1 Machine 4 Machine 3 Input 2 Output 1 Robot 2

  28. Green Production Sequence (‘safety’ specification) M1 Robot 2 O1 I1 M3 Robot 1 Robot 1 M2 Red Production Sequence (‘safety’ specification) Robot 2 Robot 1 Robot 1 I2 M4 M2 O2 ?! Robot 1 Robot 1 M4 M2 M3 Workcell Control Issues Blocking! (prohibit by nonblocking ‘liveness’ spec’n)

  29. IDD Results: Workcell

  30. Computing Effort vs. |Nodes| • Computing time ~ |Nodes|1.5 << |States| • Memory usage ~ |Nodes| K • For ‘loosely coupled’ practical systems |Nodes| ~ N  K Cwhere N = number of system components (m+n)Kstate size of individual automataC= coupling coefficient  2 • |Nodes|  linear (not exponential!)in N

  31. {0,1}n state vector   Control IDDs SUPER PLANT   new event new enabled event set Supervisor Implementation

  32. Summary • Some history • Supervisory Control Theory (SCT) • Large systems (using IDDs) • Hierarchy • Extensions and Applications • Conclusions

  33. Manager (slow) scope Operator (fast) Architecture:Hierarchical Layering • Scope  # subordinates • time horizon • bandwidth –1 • frequency –1 of significant events • Scope ratio (adjacent levels)  5:1 • e.g. 20,000 employees  7 levels

  34. ? plan = report (control  command) Hierarchical Consistency plan HI MANAGER HI WORLD advise command report fb LO OPERATOR LO WORLD control

  35. PLANThi T T*  M  (M) report  command PLANTlo   *  L  (L) report is modelled by  : L  T *, (L) =: M Report and Command -1 command is modelled by  -1 :  (M)  (L)

  36. sup M () M(M)(M)  -1  sup L() L (L)(L) Achieving Hierarchical Consistency By design of T,  arrange “ is an observer and preserves controllability” Then diagram commutes, giving hierarchical consistency

  37. M1 B1 B2 TU M2  HierarchicalTransfer Line For hierarchical control, bring in manager’s hi-level alphabet Twith events , ', ... Event  = ‘TU returns faulty workpiece for reworking’

  38.  Hierarchical Transfer Line –LO to HI

  39. fail fail pass pass SPEC - HI fail fail pass SUPER - HI pass Hierarchical Transfer Line -HI-Level Synthesis

  40. Summary • Some history • Supervisory Control Theory (SCT) • Large systems (using IDDs) • Hierarchy • Extensions and Applications • Conclusions

  41. Extensions to Base Model • Forced (preemptive) events • Timed events (delays, deadlines, forcing) - Brandin, Saadatpoor • Liveness (= eventuality), temporal logic – infinite-string ( - languages) - Fusaoka,Thistle, Ramadge • Liveness (fairness, -calculus) - Thistle,Ziller • Algebraically hybrid (?) – X = Q1  ...  Qk  n  m

  42. Some Applications • Communication protocol specification (Rudie 1990) • Rapid thermal multiprocessor (Hoffmann 1991) • Robotic agents (Kosecka 1994) • AIP automated manufacturing system • (Brandin 1994, Leduc 2001, Ma 2003) • Telephonefeature interaction (Thistle 1995) • Chemical process control • (Sanchez 1996, Alsop 1996) • Truck dispatching (Blouin 2001) • Telephone directory assistance call center (Seidl 2004)

  43. Conclusions • Achievements of SCT: * Syntheticand general * Resultscorrect by construction and computable for large systems * Modulararchitecture for management of complexity * Easy to teach and use (e.g. materials on Internet) • Challenges for SCT: • * How to interpret and modify controller structure (e.g. IDDs  linear inequalities) ? • * How to find general laws of architecture ?

More Related