Abstract: An approach for failure prediction for stochastic DESs modeled by stochastic finite state automata is proposed. Firstly, the notion of predictability of failure events of stochastic systems is formalized. Then a predictor automaton is constructed by introducing the probability transition matrix, and a necessary and sufficient condition of the predictability of stochastic systems is presented. Based on the fault predictor, a method for checking whether the considered stochastic DES has been equipped with the capability of failure prediction is developed by computing the extended Markov matrix, which results in the realization of accurate prediction of the failure events before occurring.

Key words: stochastic discrete event systems, failure diagnosis, failure prediction, Markov matrix