Fault detection and diagnosis in distributed systems: An approach by partially stochastic Petri nets (Q1273383)

For the consideration of the problem of alarm correlation in large distributed systems, a sequence of alarms is assumed. The concurrence of events is used in order to separate and simplify the state estimation in a faulty system. Petri nets and their causality semantics are used to model concurrency. Special partially stochastic Petri nets are developed; they establish a certain type of equivalence between concurrency and independence. The diagnosis problem is defined as the computation of the most likely history of the net given a sequence of observed alarms. Optimal trajectories are built by means of Viterbi-type algorithms, i.e. by connecting tiles that match the observed sequence of alarms and their causal dependence relations. It is shown that the resulting optimal trajectory, as causality graph, does not depend on which sequence of alarms is observed, provided that the same causality graph is satisfied.