On the Computation of Natural Observers for Extended Finite Automata
Paper in proceedings, 2014
Compared to finite automata, Extended Finite Automata (EFAs) allows us to efficiently represent discrete-event systems that involve non-trivial data manipulation. However, the complexity of designing supervisors for such systems is still a challenge. In our previous works, we have studied model abstraction for EFAs using natural projections with observer property on events as well as data. In this paper, we provide sufficient conditions for verifying the observer properties and further enhance the EFAs when the property does not hold. To this end, we introduce symbolic simplification techniques for data and generalize existing algorithms in the literature for the events to compute natural observers for EFAs. The importance of this combined abstraction and symbolic simplification method is demonstrated by synthesis of a nonblocking controller for an industrial manufacturing system.
supervisory control theory