On the Supervisory Control of Petri Net Behaviors

by

Professor R. S. Sreenivas

University of Illinois at Urbana-Champaign

Abstract - A large class of systems can be modeled as systems with independent, interacting, concurrent components. Typically, each independent process is split into several operations. The execution of each operation is conditioned on the satisfaction of a set of logical preconditions. Upon the execution of any such operation, a new set of logical conditions is created that inhibit the execution of some operations and enables the execution of others in the system. The supervisory control of such systems requires an external agent to regulate, or limit, the operations of each component so as to guarantee a common objective. In this talk we concern ourselves with a stronger version of deadlock avoidance called liveness. From any reachable state of a live system, it should be possible for any of the components to execute any of its operations, although not necessarily immediately. We consider systems that are modeled using Petri nets (PNs) and we present a survey of results in the literature on the existence of a supervisory policy that enforces liveness. The talk will conclude with a discussion on a few open problems.

Friday, October 28, 2005

3:30 – 4:30p.m.

Rm. 1500 EECS