HEURISTIC SUPERVISION OF DISCRETE EVENT SYSTEMS

Results on supervisory control of discrete event processes are mainly concernedwith designing a controller such that the closed loop system generates a language L whichcoincides with the target language.In this paper we consider supervisory control problem for aclass of discrete event processes when the design target is that a certain set of language patternscan not occur in the closed loop language.A heuristic supervisor is constructed via the heuristicsearch method developed in AI.This scheme requires only the information about thecontrollability of the event symbol set.It is capable of designing a modular heuristic supervisorwhen there are many such forbidden strings.Effectiveness of such a supervisor is proved forsystems with perfect and partial observation.