Enforcement of event-based supervisory constraints using state-based methods

Efficient, established techniques exist for enforcing systems of linear inequalities on the state vectors of discrete event systems modeled as Petri nets. The paper presents methods for extending the convenience of these state-based techniques to supervisory constraints that are expressed in terms of allowable plant events rather than just allowable plant states. Two interpretations of event-based constraints are described. The direct interpretation assumes that the controller will actively disable events that would directly violate the constraints. The indirect interpretation indicates that the controller should prevent those states that could lead to the violation of the constraints. Methods for automatically synthesizing supervisors that use either of these interpretations are presented here along with mathematical results for integrating these methods with established techniques for handling uncontrollable and unobservable plant transitions. The technique is illustrated with a process control example.