44 / 2025-03-28 19:18:58

Remaining Useful Life Prediction of Rolling Bearings Based on Full-Band Feature Feedback Adaptive Fusion

Rolling bearings,Remaining useful life,Full-frequency-band feature extraction,Wasserstein distance,Adaptive feedback fusion

Rolling bearings serve as the core components in mechanical equipment, and accurately predicting their Remaining Useful Life (RUL) is crucial for ensuring equipment reliability and reducing maintenance costs. However, existing RUL prediction methods often underutilize frequency-domain features and fail to account for prediction outcomes in feature fusion. To address these issues, this paper proposes an RUL prediction framework based on full-frequency-band degradation feature extraction and adaptive feedback fusion. First, a 1/3 binary tree filtering strategy is used to perform multi-scale decomposition of the vibration signals, and the Wasserstein distance (WD) is employed to extract full-frequency-band degradation features. Next, a dynamic feedback mechanism is introduced to adaptively adjust Health Indicators (HI). Experimental results on the XJTU-SY dataset confirm the effectiveness of the proposed method