Inverter overcurrent is one of the main faults in the traction system of Shenhua twelve-axle electric locomotive, which involves dozen fault causes that have been analyzed and determined, and in different working conditions, there are hundreds of fault characteristic change caused by different reasons, which lead to difficulties of analyzing and dealing with the faults. Fault feature variables were extracted according to expert experience, and fault was located based on the application of data mining and characteristics of statistics analysis method, which improved the accuracy and robustness of fault diagnosis.