Abstract: The predicting ability of the fault prediction algorithm based on SIR particle filter will be badly influenced by sample impoverishment,which is one of the main disadvantages for the application of particle filter in fault prediction.A fault prediction algorithm based on weight selected particle filter is proposed to resolve the above problem.According to their weights,the better particles are selected from a vast amount of particles to improve the diversity of samples.As a result,the problem of sample impoverishment is ameliorated and the tracking ability of the fault prediction algorithm based on SIR particle filter is improved.Simulation results demonstrate that the fault prediction algorithm based on weight selected particle filter is feasible.