Supervisory control of deterministic Petri nets with regular specification languages

Algorithms for computing a minimally restrictive control in the context of supervisory control of discrete-event systems have been well developed when both the plant and the desired behaviour are given as regular languages. In this paper the authors extend such prior results by presenting an algorithm for computing a minimally restrictive control when the plant behaviour is a deterministic Petri net language and the desired behaviour is a regular language. As part of the development of the algorithm, the authors establish the following results that are of independent interest: i) the problem of determining whether a given deterministic Petri net language is controllable with respect to another deterministic Petri net language is reducible to a reachability problem of Petri nets and ii) the problem of synthesizing the minimally restrictive supervisor so that the controlled system generates the supremal controllable sublanguage is reducible to a forbidden marking problem. In particular, the authors can directly identify the set of forbidden markings without having to construct any reachability tree.