Fault-diagnosis in discrete event systems: Improvements and new results

Abstract

The malfunction of sensors, actuators, and erroneous actions of human operators can have some disastrous consequences in high risk systems especially if these systems have multiple faults that can lead to undesirable shutdowns and consequently mass reduction. A reduced interpreted Petri net (IPN) diagnoser has been devised only for safe Petri net models with an output function that associates an output vector to each net marking. The main drawback of this approach is that the Petri net model of the system to be monitored should be diagnosable i.e. all faults can be detected that limits its application on a set of diagnosable models. For non diagnosable Petri net model, the conventional diagnoser incidence matrix has columns with null or similar values that fail to detect a single fault. The conventional diagnoser also cannot detect multiple faults even for diagnosable models. This paper introduces a new local diagnoser to overcome such problems. It decomposes the central IPN-diagnoser into a set of local diagnosers that are linked with multi sessions of the process to be monitored. This decomposition should guarantee that the developed local diagnosers have incidence matrices that their columns are different from each other. For null values contained in the incidence matrix of a local diagnoser, this paper proposes a set of rules based on the synchronic composition idea to overcome this problem. This proposed scheme allows multiple faults detection and isolation in quick and accurate manner for all Petri net models. Industrial processes are employed for testing the soundness of the proposed scheme.