Rolling bearing abnormal wear intelligent monitoring and fault diagnosis based on multi-source information fusion with oil debris

Abstract Rolling bearings in aero-engine rotor systems are prone to wear failure due to installation deviations and manufacturing defects. The rapid deterioration of bearing wear can significantly compromise the safety and reliability of the rotor system and the entire machine. Accurate online monitoring of the bearing wear state is crucial to ensuring the safe and stable operation of the rotor-bearing system. However, in rolling bearing condition monitoring, the state information provided by a single type of data is often limited and insufficient to accurately reflect the bearing fault state. Multi-source information, particularly oil metal debris signals, can be utilized to achieve more precise monitoring and assessment of the bearing condition. This paper investigates rolling bearing wear faults and their long-term operational evolution through experiments conducted on a custom rotor-bearing test rig. A multi-physical parameter monitoring approach is employed to analyze collected data, including bearing vibration, temperature, and lubricating oil debris. First, an online monitoring method for bearing wear state is proposed, utilizing oil debris analysis and a Kalman filter. Next, a multi-source fault feature fusion method is adopted, integrating vibration features, particle features, and temperature features. Principal component analysis is applied to extract key sensitive feature parameters from the feature parameter set. Finally, the factors influencing the accuracy of bearing wear fault diagnosis are analyzed and compared using SVM, KNN, and decision tree methods. This analysis determines the relative importance of debris and temperature in bearing wear diagnosis, offering a novel reference for fault detection.