Comprehensive Remaining Useful Life Prediction for Rolling Element Bearings Based on Time-Varying Particle Filtering

The remaining useful life (RUL) prediction of rolling element bearing has attracted substantial attention due to its importance in improving reliability and safety of machines. Particle filter (PF) algorithm is widely utilized to track the degradation process of mechanical equipment in RUL prediction. In engineering applications, due to the poor consistency of bearings, the classical PF model with a certain model finds it difficult to deal with bearings with different degradation trends, resulting in poor generalization ability. Therefore, a time-varying PF (TVPF)-based comprehensive RUL (CRUL) prediction model is developed in this article. In the model, a TVPF algorithm is constructed by an adaptive selection rule and sliding window, it has the capability to select the optimal state model with a sliding window, according to the characteristics of the data, and track the degradation state of bearings with different degradation trends; meanwhile, a global/local information fusion (GLIF) technique is proposed for comprehensively considering the overall information and the latest degraded state of the rolling bearings. The effectiveness of the proposed method is verified by two datasets with different degradation trends, respectively. The comparative study indicates that the proposed TVPF algorithm outperforms the other state-of-art methods in RUL prediction and system prognosis with respect to better accuracy and robustness.