On use of Petri-nets for diagnosing nonpermanent failures

Several frameworks namely finite state machines, process algebra, Petri nets (PN) etc. have been used for failure modeling and diagnosability analysis in discrete event systems. Many systems exhibit nonpermanent failures in the sense that the system recovers to normal condition after failure. Diagnosability conditions, proposed in the literature, are mainly for permanent failures. In this paper failure diagnosability for nonpermanent failures has been studied for PN based models. In this paper it is first shown using an example that a new diagnosability condition is required for handling nonpermanent failures. Following that a new diagnoser condition is achieved for nonpermanent failures and shown to be necessary and sufficient for PN based models.