Fault probability evaluation in DESs with an algorithm of linear complexity and partially observed SPNs

This article concerns fault diagnosis for stochastic discrete event systems. For this purpose, partially observed stochastic Petri nets are introduced that include (i) the Markovian stochastic dynamics used to represent failure processes and also (ii) the model of the sensors used to measure the events and markings. Timed observation sequences result from this modeling and the probabilities of marking trajectories consistent with a given timed observation sequence are systematically computed. Diagnosis in terms of faults probability is obtained as a consequence and an algorithm of linear complexity is proposed to evaluate the fault probability on the basis of segments extracted from the timed observation sequence. Performance of the algorithm is discussed wrt to usual non detection and false alarm rates.