A survey on efficient diagnosability tests for automata and bounded Petri nets

This paper presents a survey and evaluation of the efficiency of polynomial diagnosability algorithms for systems modeled by Petri nets and automata. A modified verification algorithm that reduces the state space by exploiting symmetry and abstracting unobservable transitions is also proposed. We show the importance of minimal explanations on the performance of diagnosability verifiers. Different verifiers are compared in terms of state space and elapsed time. It is shown that the minimal explanation notion involved in the modified basis reachability graph, a graph presented by Cabasino et al. [3] for diagnosability analysis of Petri nets, has great impact also on automata-based diagnosability methods. The evaluation often shows improved computation times of a factor 1000 or more when the concept of minimal explanation is included in the computation.