Current-Based Gear Fault Detection for Locomotive Gearboxes

Our objective is to analyze the traction motor stator current in the diagnosis of the gearbox fault of the locomotive. The gear localized defects cause torque oscillations which modulate the frequency components in the stator current signal and generate the sidebands in the stator current power spectrum. However, the sideband components are weak and covered by noise. Dual-tree complex wavelet transform is used to extract fault features hidden in the sidebands. We use wavelet energy entropy to make a preliminary diagnosis of the signal, and then apply a dual-tree complex wavelet transform in the current signal to further determine the type of failure. In order to illustrate the validity of the proposed diagnosis method, simulation signal analysis was carried out and then verified in the laboratory. Finally, gear diagnosis experiments on the freight locomotive show that the proposed method can effectively diagnose gear failures.