A Method for Fault Diagnosis of Train Axle Box Bearings Based on Impact Detection

To address the issues of poor decoupling ability and stability in traditional impact pulse methods, a train axle box bearing fault diagnosis method based on impact detection is proposed. Firstly, fast squared envelope kurtogram (FSEK) analysis, resonance demodulation and adaptive frequency band search are utilized to extract the target impact components from the vibration signals. Then, a high fault sensitivity inverse standard deviation (ISD) indicator is constructed, which can effectively extract and amplify the impact characteristics of the fault signal. Finally, a regional averaging strategy is used to achieve continuous and automatic fault detection for long time series data. The effectiveness, stability, and decoupling performance of the proposed indicator and method are verified using test bench data and real vehicle data. The results show that the proposed indicator can effectively extract the impact characteristics in fault signals. Compared with other methods, the proposed method demonstrates better stability under varying operating conditions and superior decoupling performance under multi-source vibration coupling. In the test with bench test data, both the accuracy and recall rate of the proposed method reached 100%.