Abstract: To solve the problems of the conventional spectral analysis method in train bearing fault diagnosis that always lead to misdiagnosis and missed diagnosis,the paper analyzes the vibration mechanism of the train bearing spectrum-varying fault,and proposes a diagnosis method based on data interpolation re-sampling analysis.The train bearing's vibration integral periodic and equal angle signal is collected through the pulse-triggered sampling method which is researched in this paper.Through the data interpolation and characteristic spectral identification method,the train's vibration frequency spectrum-varying fault is easy to be diagnosed.Simulation analysis and experimental results show that this diagnostic method can accurately and effectively identify the train early bearing fault.