A fault detection and diagnosis approach based on hidden Markov chain model

A fault detection and diagnosis (FDD) approach based on a hidden Markov chain model is proposed. In the proposed approach, the occurrence or recovery of a failure in a dynamic system is modeled as a finite-state Markov (or semi-Markov) chain with known transition probabilities. For such a hybrid system, either the interacting multiple-model (IMM) or the first-order generalized pseudo-Bayesian (GPB1) estimation algorithm can be used for state estimation, fault detection and diagnosis. The superiority of the approach is illustrated by an aircraft example for sensors and actuators failures. Both deterministic and random fault scenarios are designed and used for evaluating and comparing the performance. Some performance indices are presented. The robustness of the proposed approach to the design of model transition probabilities, fault modeling errors, and the uncertainties of noise statistics are also evaluated