Modular Fault Diagnosis Based on Discrete Event Systems

Abstract

An awareness of failure type and location is an indispensable requirement for the establishment of adequate recovery strategies and the maintenance of Factory Automation and Process Control systems. The failure diagnosis methodology presented in this paper is based on Discrete Event Systems models and in the diagnoser concept, which enable the off-line and on-line analysis of systems failures. We present an approach for models and associated diagnosers based on a modular decomposition of the global system, with the aim of avoiding problems of exponential explosion in the number of states and computational complexity of the diagnosis process.