Application of WOA-VMD-SVM in Fault Diagnosis of Generator Inter-turn Short Circuit

Abstract

This paper proposes a feature extraction method based on whale optimization algorithm and variational mode decomposition (WOA-VMD) to overcome the low feature extraction accuracy of generator early inter-turn short circuit fault. WOA-VMD process the current signal, and the sample entropy is taken as the fitness function of WOA to optimize the VMD parameter combination of modal components' number K and penalty parament α. Then, the optimized VMD decomposes current signals into K intrinsic mode functions (IMFs). IMFs with higher kurtosis values are selected to extract energy entropy as the feature vectors. Finally, the whale optimization algorithm and support vector machine (WOA-SVM) pattern recognition model is used to classify the feature vectors and diagnose generator inter-turn short circuit degree. The experiments show that the proposed method extracts the weak fault features in the early inter-turn short circuit signal and improves the fault diagnosis accuracy, reaching 97.75%.