137 / 2025-05-26 16:00:49

Research on Bearing Fault Diagnosis Technology Based on Multi-Source Data Fusion

multi-source data fusion, bearing fault diagnosis, intelligent maintenance

Bearings, as the core components of industrial equipment, often operate under prolonged high loads, which makes them susceptible to failures. Therefore, there is a need for health monitoring and fault diagnosis to ensure operational safety. This paper first reviews and summarizes mainstream traditional fault diagnosis approaches in the domains of signal processing, physical modelling, and data-driven methods, and highlights their inherent limitations. It then focuses on multi-source data fusion techniques, with particular attention to three fusion architectures: data-level, feature-level, and decision-level. The advantages of these approaches are demonstrated through typical application cases such as vibration signal stacking and time–frequency image fusion.In the section on compound bearing fault diagnosis methods, current research on multi-source data fusion is categorized into three innovative strategies: collaborative data fusion, deep feature extraction, and dynamic weight optimization, based on different underlying principles. A comparative analysis is conducted to discuss the focus, application domains, and strengths and weaknesses of each method. Looking ahead, future research in bearing fault diagnosis based on multi-source data fusion should explore concurrent fault collaboration, lightweight architecture design, interpretable model development, and distributed processing techniques, in order to overcome the limitations of existing approaches and enhance the reliability of bearing fault diagnosis technologies.