A hidden Markov model-based algorithm for fault diagnosis with partial andimperfect tests

ln this paper, we present a hidden Markov model (HMM) based algorithm for f ault diagnosis in systems with partial and imperfect tests. The HMM-based a lgorithm Ends the most likely state evolution, given a sequence of uncertai n test outcomes over time. We also present a method to estimate online the HMM parameters, namely, the state transition probabilities, the instantaneo us probabilities of test outcomes given the system state and the initial st ate distribution, that are fundamental to HMM-based adaptive fault diagnosi s, The efficacy of parameter estimation method is demonstrated by comparing the diagnostic accuracies of an algorithm with complete knowledge of HMM p arameters with those of an adaptive one. In addition, the advantages of usi ng the HMM approach over a Hamming-distance based fault diagnosis technique are quantified. Tradeoffs in computational complexity versus performance o f the diagnostic algorithm are also discussed.