Diagnosis with Causality Relationships and Directed Paths in PN Models

Authors:	Lefebvre Dimitri, University Le Havre, France Delherm Catherine, University Le Havre, France
Topic:	5.2 Manufacturing Modelling for Management and Control
Session:	Manufacturing Modelling, Management and Control
Keywords:	Petri nets, structural properties, sensitivity analysis, estimation, diagnosis.

Abstract

Petri nets are a suitable tool for the diagnosis of discrete event systems. For this purpose, faulty behaviours are modelled by the firing of failure transitions. This paper is about structural sensitivity in Petri net with respect to the firing of the failure transitions. Algebraic results are provided to characterise the influence and dependence areas of the failure transitions and diagnosability of the systems is obtained as a consequence. The main advantage of our approach is to investigate the diagnosability without working out the marking tree of the diagnoser.