An integrated dual-scale similarity-based method for bearing remaining useful life prediction

As a pivotal technology of Prognostic and Health Management, the remaining useful life (RUL) prediction techniques significantly contribute to predictive maintenance and ensure the safe operation of mechanical equipment. Nevertheless, the current similarity-based prediction (SBP) methods face challenges in effectively utilizing the degradation information encapsulated within a limited number of degradation samples. Therefore, an integrated dual-scale similarity-based prediction (IDS-SBP) method is proposed bearing RUL prediction, which can fully mine the degradation information of the samples from two distinct time scales. Specifically, a whole lifecycle dynamic model is constructed to describe the various long-term degradation processes for bearings, which enriches the variety of the performance degradation samples. Subsequently, the dual-scale matching strategy is designed to extract the degradation information from two different time scales. Meanwhile, the designed lifetime calibration technique can calibrate the lifetime of samples by considering the degradation rate. Finally, the uncertainty analysis is conducted to integrate the prediction results at different time scales, thereby achieving the comprehensive evaluation of test bearings. Several sets of experimental data are applied to verify the prediction performance of the proposed method, and prediction results confirm that the proposed method achieves great prediction accuracy and superior generalization ability.