Vibration nonstationary representation of the railway vehicle's gear system considering spalling failure under variable speed conditions: Numerical modeling and field testing

The nonstationary representation of spalling faults in the helical gear transmission system of railway vehicles remains unclear under variable speed conditions, and the selection of sensitive features in faults recognition has been confirmed as a problem to be addressed in this field. Thus, a vehicle-track coupled model with a helical gear transmission system was built in this study, which considered the spalling faults and employed field testing data to verify the model. The model was adopted to analyze the non-stationary representation of a helical gear system with spalling faults in the translational directions under variable speed conditions. Subsequently, the data originating from the simulation and field testing were adopted in this study, such that the ReliefF-MSVM (Multi-class SVM) algorithm was driven to sift the sensitive features in the translational directions and obtain the optimal sensitive features and diagnostic direction. The result achieved with the sensitive features selected suggested that the vertical statistical characterization of the helical gear transmission system of railway vehicles exhibited low dimensionality and high robustness, and the diagnostic results were optimal. This study can provide a theoretical reference for dynamics analysis, health monitoring, and fault diagnosis in the helical gear system of railway vehicles.