Distributed identification of concurrent discrete event systems for fault detection purposes

An identification method for concurrent discrete event systems (DES) is presented. The purpose of the identified model is model-based fault detection in closed loop industrial DES such as production or manufacturing systems. The method consists of identifying an automata network with a scalable restriction of the network behavior. We especially deal with incomplete observation of the system language within a finite time horizon which is a typical result of a high degree of concurrency. It is shown how models identified on the basis of an incompletely observed language can be used for fault detection.