Application of multi-class fuzzy support vector machine classifier for fault diagnosis of wind turbine

This paper presents an approach for fault diagnosis of wind turbine (WT) based on multi-class fuzzy support vector machine (FSVM) classifier. In this method, empirical mode decomposition is adopted to extract fault feature vectors from vibration signals. FSVM is used for solving classification problem with outliers or noises, where kernel fuzzy c-means clustering algorithm and particle swarm optimization algorithm are applied to calculate fuzzy membership and optimize the parameters of kernel function of FSVM, respectively. In addition, to study the performance of the proposed approach, another two widely used methods, named back propagation neural network and standard support vector machine, are studied and compared. Discrete wavelet transform is also used to extract fault feature vectors. To validate the proposed approach, a direct-drive WT test rig is constructed and the experiments are carried out. The experimental results show that the proposed approach is an effective fault diagnosis method for WT, which has a better performance and can achieve higher diagnostic accuracy. A multi-class FSVM classifier constructed by one-against-other method is used for fault diagnosis of wind turbine.KFCM algorithm is extended to calculate fuzzy membership values of training samples for a multi-class FSVM classifier.PSO algorithm is applied to optimize the parameters of the kernel function of FSVM.Fault diagnosis steps are presented, and the effectiveness of the proposed method is validated.