Abstract: This paper analyzes the working and fault principles of the converter,which is the key-parts of SS7E locomotive,and classifies the converter faults.A method based on wavelet analysis is presented to extract energy eigenvector from the output voltage wave of SS7E locomotive converter,and a diagnosis scheme based on energy eigenvector and neural-network is proposed for converter fault diagnosis.The hardware frame and software design of the fault diagnosis system are discussed and the application scheme is proposed to diagnose the specific position of converter fault online by increasing a few monitoring points.The validity of the presented method is proved with computer simulation.