Structural Diagnosability of DES and Design of Reduced Petri Net Diagnosers

This paper deals with diagnosis of permanent and operational faults of partially observed discrete event systems modeled by interpreted Petri nets capturing both normal and faulty behaviors. Two main results are presented: a structural characterization of the diagnosability property and a method for designing reduced model diagnosers for online fault detection and location. Sufficient conditions for diagnosability are provided based on the analysis of the influence area of every fault
$f_{i}$
in the model and the relative distance between pairs of transitions; polynomial algorithms are proposed for determining diagnosability. The diagnoser includes two reduced models that monitor the system; one for tracking the actual behavior and the other for establishing the expected behavior; the difference of markings in such models, called residue, provides enough information for the immediate location of faults, even if they occur simultaneously.