Abstract: To solve certain kinds of fault prognostic problems, an algorithm based on particle filter is presented. At the state estimation stage, the algorithm estimates the posterior distribution of the states and parameters of the system fault progression model based on hybrid system particle filter and dual estimation. At the state prediction stage, the algorithm converts the problem of predicting the continuous states of a hybrid system model to the problem of predicting the states of a basic state space model under certain predefined assumptions. By sampling iteratively the posterior distribution of current continuous states, the algorithm can use the sampled particles to form the state prior distribution for some future time. At the prognostic decision stage, based upon the above calculated continuous state distribution, combined with certain fault criteria, the distribution of system remaining useful lifetime can then be inferred. Simulation result demonstrates the validity and feasibility of the proposed algorithm.

Key words: fault prognostics, stochastic systems, hybrid system particle filter, dual estimation, sampling importance resampling, distribution of remaining useful lifetime