A Structural Characterization of Diagnosable Petri Net Models

This paper is concerned with the diagnosability property of discrete event systems (DES) modeled by live and safe Interpreted Petri nets (IPN). The IPN are used to model both, the normal and faulty behavior of the system. Based on this model, the notion of input-output diagnosability is introduced and a polynomial algorithm to characterize diagnosable IPN is proposed. The novel features of the approach herein presented are: a) the notion of relative distance between any pair of transitions; b) the use of the net siphons and T-semiflows to determine the relative distance between any pair of transitions; c) a characterization of diagnosable IPN based on the relative distance concept. Moreover, the approach herein presented characterizes a broader class of IPN exhibiting the diagnosability property and presents a better deepening of the structures needed to characterize diagnosable IPN.