Abstract: Local fluctuations in the degradation critical region of mechanical equipment can lead to the failure of the degradation thresholds, bringing challenges to the diagnostic and maintenance. In high-precision mechanical equipment, the elimination of such uncertainty fluctuations also results in the loss of abundant fault information in the degradation critical region, which in turn affects the accuracy of the degradation model. In view of this, the concept and methodology of multiscale degradation modeling based on localized fluctuations is proposed. By intricately integrating vibration signals across different temporal scales with physical models of failure evolution at various spatial scales, a multiscale model for equipment performance degradation assessment is developed. The results transform unfavorable local fluctuations into monotonically stable degradation assessment metrics under synergistic temporal-spatial scale modeling, which not only avoids failure of degradation thresholds due to uncertainty fluctuations, but also provides degradation features with training value, further highlighting the validity of evaluating the degradation of physical system performance at the temporal-spatial scale. Taking the degradation process of rolling contact fatigue failure as an illustration, the degradation differences of plastic and brittle materials containing steel and ceramics under rolling contact fatigue are discussed. The effectiveness and robustness of the degradation indices obtained through the proposed method are validated across diverse application scenarios.

Key words: failure of degradation threshold, local fluctuations, degradation modeling, performance assessment