IGBT Open Circuit Fault Diagnosis Based on Improved Support Vector Machine

Modular multilevel converter (MMC) is a new type of the voltage source converter, which is widely used in the flexible DC transmission and motor drive. However, the MMC is composed of a large number of sub-modules, which poses a huge difficulty for accurately locating the specific sub-module that has a fault. Therefore, this paper proposes an improved support vector machine (SVM) based on the overlapped wavelet packet transform (MODWPT) to diagnose the open circuit fault of the insulated gate bipolar transistor (IGBT) of the MMC sub-module. The MODWPT is used for the feature extraction, then the k-fold cross-validation can group fault feature data sets to evaluate the performance of SVM classifiers, which can effectively reduce the generalization error of the fault diagnosis model. Based on the MMC fault simulation model of the PSCAD platform, the experimental results show that the average fault diagnosis accuracy of the improved SVM based on the MODWPT is 99.78%, which has better classification accuracy and reliability than the traditional SVM, the back propagation neural network and Bayesian.