In: Proceedings of the 24th International Conference on Applications and Theory of Petri Nets (ICATPN 2003), Eindhoven, The Netherlands, June 23-27, 2003, pages 316-336. Volume 2679 of Lecture Notes in Computer Science / Wil M. P. van der Aalst and Eike Best (Eds.) --- Springer-Verlag, June 2003.
Abstract: The problem of detecting and isolating fault events in dynamic systems modeled as discrete-event systems is considered. The modeling formalism adopted is that of Petri nets with labeled transitions, where some of the transitions are labeled by different types of unobservable fault events. The Diagnoser Approach for discrete-event systems modeled by automata developed in earlier work is adapted and extended to on-line fault diagnosis of systems modeled by Petri nets, resulting in a centralized diagnosis algorithm based on the notion of "Petri net diagnosers". A distributed version of this centralized algorithm is also presented. This distributed version assumes that the Petri net model of the system can be decomposed into two place-bordered Petri nets satisfying certain conditions and that the two resulting Petri net diagnosers can exchange messages upon the occurrence of observable events. It is shown that this distributed algorithm is correct in the sense that it recovers the same diagnostic information as the centralized algorithm. The distributed algorithm provides an approach for tackling fault diagnosis of large complex systems.