A stochastic automaton approach to discriminate intermittent from permanent faults

Abstract

The conventional model-based diagnosis usually potentially presumes faults are persistent and does not take intermittent faults into account, which is the major cause of the problems of false alarms, cannot duplicate and no fault found in aircraft avionics and present a tremendous challenge to prognostics and health management. Aiming at the problem that the logical automaton proposed by Sampath et al. cannot distinguish between strings or states that are highly probable and those that are less probable, a stochastic automaton approach is given to distinguish the fault types by extending the fault model to include both permanent faults and intermittent faults. The notions of A- and AA-diagnosability of permanent faults and intermittent faults for stochastic automaton are defined. Thereafter, the diagnoser with a probability matrix appended to each transition that can be used to update the probability distribution on the state estimate is constructed. Finally, an example of aeronautic gyroscope is presented to demonstrate the proposed approach, and the analysis results show that this approach is able to discriminate the fault types within bounded delay if the system is A- and AA-diagnosable. In our previous paper, we have extended the logical automaton model, and investigated the stochastic automaton approach in this article.