ABSTRACT
This chapter presents the salient features of a general methodology for fault diagnosis in discrete event systems modelled as partially observed finite-state automata. The methodology presented in this chapter is generic and applicable to any discrete event system modelled as an automaton. It discusses the manufacturing examples and in particular discusses the modelling of automated manufacturing systems composed of interacting sensors, actuators and physical devices, coupled by control logic. The framework of discrete event system theory is able to address the notion of events being unob-servable. An event being unobservable implies that physically there is no sensor, or algorithm, for directly detecting the occurrence of that event. The chapter illustrates the complete process of model building, controller design, diagnoser construction and diagnosability analysis, in the context of a simple component found in automated manufacturing systems, a single acting and double feedback pneumatic device.
