The Svm-based Machine Owners Converter Fault Diagnosis

As we all know, security is the lifeline of railway transportation. The reliable running of locomotive is the basic protection of railway safe production. The real-time detection and fault diagnosis of locomotive is an important technical measure to prevent failure, ensure the safe running, and improve the efficiency of transportation.The study project in this paper came from the National 863 Program "train security status monitoring and fault warning technology research" and the National Natural Science Foundation Project "Fault Diagnosis of high-speed train system based on the Train Communication Network".First of all, the significance of the fault diagnosis to locomotive equipment was clarified. Then, the development of fault diagnosis to locomotive was presented and the SVM was introduced, which is a fresh theory in the mode identification and machine learning of data mining field. A new method based on SVM was applied in the fault diagnosis of the electric locomotive converter.The electric locomotive converter was chosen to be researched, whose circuit structure and running principles ware introduced. After analyzing the mechanism of the converter, the distribution of circuit failure was analyzed and the output voltage was chose as the judge basis for fault diagnosis .With the Simulink toolbox in Matlab, the output voltages of different faults ware obtained. Making use of wavelet analysis to process the output voltage, extractted the fault feature and construct the feature vector. The samples which contain the fault information ware accessed finally. With the SVM toolbox, the multi-fault classifiers ware established. However, in the infinite space, SVM easily leads to wrong classification. For this problem, an improved method was proposed in this paper, which can make SVM to classify in a limited space. This is the innovation of this article. Finally, the simulation samples ware trained and tested. Experimental results showed that the classification result was fine and achieved the desired objective. In short, the diagnosis method proposed in this paper was real-time, high accuracy and was a positive theory guidance for the fault diagnosis of Electric locomotive converter.