Failure diagnosis of dynamic systems: an approach based on discrete event systems

We present the salient features of a methodology for failure diagnosis of dynamic systems that can be modeled as discrete event systems. This methodology was introduced by Sampath et al. for centralized systems and subsequently extended by Debouk et al. (2000) for certain classes of decentralized systems. We discuss how to perform detection and identification of unobservable fault events using diagnosers, which are finite-state automata that are built from the discrete-event model of the system under consideration. Examples of diagnosers are given. Comparisons with other methodologies for diagnosing dynamic systems are given.