Multiple Fault Diagnosis for Three-level Inverter Via Wavelet Analysis and Support Vector Machine

Abstract

Inverter is an important part in the locomotive traction system. In this paper a fault diagnosis method of three-level NPC inverter with two IGBT failures concurrently in different bridges was introduced. With the Simulink toolbox in Matlab, We obtained the output voltages’ difference between faulty and fault-free cases, get ï„Uaã€ï„Ubã€ï„Uc , d-q transformation on ï„Uaã€ï„Ubã€ï„Uc , put three-phase into two-phase, Ud and Uq. Making use of wavelet analysis to process Ud and Uq, extracted the fault feature and construct the feature vector. The samples which contain the fault information were accessed finally. With the LIBSVM toolbox, the multi-fault classifiers were established. Finally, the simulation samples were trained and tested. Experimental results showed that the classification result was fine and achieved the desired objective.