Supervisory design using Petri nets

Abstract

A Petri net model is described that is consistent with the control theory for discrete event systems. The purpose is to apply the theoretical approach of supervisory control to a model that permits an efficient solution of the control problem. The focus is at the model level. The authors show how Petri nets may be used to design a supervisor. The design requires two steps. In the first step, a coarse structure for a supervisor is synthesized by means of a concurrent composition of different modules. In the second step, the structure is refined to avoid reaching forbidden markings. The refinement procedure may always be applied when the net is conservative. In both steps, the use of Petri nets allows the structure of the model to be small.<>