Research on Rolling Bearing Fault Diagnosis Based on Multiscale Hybrid-Domain Information Entropy

To enhance the accuracy and reliability of fault diagnosis for rolling bearings in aero-engines and address the poor stability of traditional time-domain feature parameters under varying operating conditions, this paper proposes a fault feature extraction method based on multiscale hybrid-domain information entropy. Utilizing Shannon entropy theory and the concept of information fusion, this method integrates information entropy from the time-domain, frequency-domain, and time-frequency domain to construct a comprehensive feature vector. Specifically, it employs entropy measures such as fuzzy entropy, power spectrum entropy, and singular spectrum entropy, and calculates the hybrid-domain information entropy matrix using normalized Euclidean distance. The method is experimentally validated on multiple bearing fault types, analyzing entropy characteristics under different fault conditions. Experimental results demonstrate that the proposed method exhibits good stability and robustness across various fault conditions. The hybrid-domain information entropy fusion method significantly improves fault diagnosis accuracy and consistency, effectively extracting fault features of rolling bearings. This proposed method is not only applicable to fault diagnosis of aero-engine rolling bearings but also holds broad application potential for health monitoring and fault prediction in other complex mechanical systems. It introduces new perspectives and tools for fault diagnosis in complex mechanical systems, contributing to enhanced precision and reliability in equipment health management.