120 likes | 241 Vues
This discussion explores the intricacies of supervisory control by solving automata equations, focusing on standard realizations, largest solutions, and progressive solutions. It covers topics such as partial controllability, observability, and solvability equations. The approach highlights the polynomial complexity and the potential for finding optimal supervisors from the largest solution. The automata equation method is shown to be applicable in dealing with various topologies, offering insights for future research in this field.
E N D
Discussion on Supervisory Control by Solving Automata Equation Victor Bushkov, Nina Yevtushenko, Tiziano Villa, Tomsk State University (Russia), University of Verona (Italy).
Problem S S P P C ? P XS C
Supervisor Example Standard realizationInit(S) – prefix-closure of S Plant P Specification S Standard realization of supervisor C Init(S)
Largest Solution Largest solution (P S)pref of P XS is a solution which includes every possible solution of P XS. Largest solution Plant P Specification S C (P S)pref
Progressive Solution Solution C of P XS is a progressivesolution if in P Ca final state could be reached from every state Non-progressive solution Cnon-prog Progressive solution Cprog Plant P P Cnon-prog P Cprog 2 Specification S
Partial Controllability • =c uc • uc-extensionCucofCis obtained by adding self-loops under everya ucat every statecofCif there are no transitions fromcunder a Plant P , uc = {c} Largest solution C (P S)pref Largest solution under partial controllability C C Cuc Specification S
Partial Controllability: Solvable Equations Supervisory Control Automata Equations 1. Init(L(S))ucInit(L(P))Init(L(S)) 2. L(S) = Init(L(S)) L(P) Init(L(S))(uc)* L(P)L(S) L(S), L(P) – languages ofS andP Init(L(S)) – prefix-closure ofL(S) Plant P , uc = {a} Specification S
Partial Controllability: Progressive Solutions Solution C of P XS is a progressivesolution under partial controllability if Cuc is a progressive solutionofP XS Plant P , uc={a} Solution under partial controllability C Cuc P Specification S
Partial Observability • =o uo • uo-foldingCuoofCis obtained by replacing each transition at every statecofCunder a uoby a self-loop C1 C2 Solution under partial observability C1 Plant P , uo = {a} Solution under partial observability C2 (C1 C2)uo Specification S
Conclusions and Future Works • The complexity of solving supervisory control problem by automata equation approach is polynomial • By solving automata equation we can find a largest supervisor, from which an optimal supervisor can be extracted • The automata equation approach can deal with more general topologies
Thank you for yourattention Questions?