Fault Diagnosis in Discrete Event Systems Using Interpreted Petri Nets

Abstract

Diagnosability property and fault detection schemes have been widely addressed on centralized approaches using the global model of the Discrete Event System (DES). Roughly speaking, diagnosability is the property of determining if using the system model is possible to detect and locate the faulty states in a finite number of steps. In the works (Sampath, et al., 1995) and (Sampath, et al., 1996), a method for modeling a DES using finite automata is proposed; based on this model, a diagnoser is derived. The cycles in the diagnoser are used to determine when the DES is diagnosable. Recently, fault diagnosis of DES has been addressed through a distributed approach allowing breaking down the complexity when dealing with large and complex systems (Benveniste, et al., 2003; O. Contant, et al., 2004; Debouk, et al., 2000; Genc & Lafortune, 2003; Jiroveanu & Boel, 2003; Pencole, 2004; Aramburo-Lizarraga, et al., 2005). In (Debouk, et al., 2000) it is proposed a decentralized and modular approach to perform failure diagnosis based on Sampath's results (Sampath, et al., 1995). In (Contant, et al., 2004) and (Pencole, 2004) the authors presented incremental algorithms to perform diagnosability analysis based on (Sampath, et al., 1995) in a distributed way; they consider systems whose components evolve by the occurrence of events; the parallel composition leads to a complete system model intractable. In (Genc & Lafortune, 2003) it is proposed a method that handles the reachability graph of the PN model in order to perform the analysis similarly to (Sampath, et al., 1995); based on design considerations the model is partitioned into two labelled PN and it is proven that the distributed diagnosis is equivalent to the centralized diagnosis; later, (Genc & Lafortune, 2005) extend the results to systems modeled by several labelled PN that share places, and present an algorithm to determine distributed diagnosis. In (Qiu & Kumar, 2005) it is studied the codiagnosability property, this property guarantees that any faults occurred in the system must be detected by at least one local diagnoser in a finite number of steps using the local information, besides, a notion of safe-codiagnosability is mentioned to capture the fact that the system has a safe specification while the system performance is tolerable. (Aramburo-Lizarraga, et al., 2005) proposes a methodology for designing reduced diagnosers and presents an algorithm to split a global model into a set of communicating sub-models for building distributed diagnosers. The diagnosers handle a system sub-model and every diagnoser has a set of communication events for detecting and O pe n A cc es s D at ab as e w w w .ite ch on lin e. co m