Wavelet and Hilbert Huang transform based wind turbine imbalance fault classification model using <i>k</i>-nearest neighbour algorithm

Abstract

Wind turbine (WT) is a key part in wind power generator system (WTGs). For proper operation, condition monitoring and fault diagnosis is a major part in WTGs. In this paper, three different types of nacelle yaw faults along with healthy condition are analysed using wavelet transform (WT) and Hilbert Huang transformed (HHT) based k-nearest neighbour (k-NN) algorithm. For decomposing the raw signals, discrete approximation of Meyer wavelet function (DMeyer/dmey) is used and to extract the feature, Hilbert Huang transform is used to find the amplitude and phase feature of decomposed signal. k-nearest neighbour algorithm based classifier is designed for classifying faults based on extracted features. Prepared feature matrix of twenty one attributes is used for wind turbine nacelle yaw imbalance fault classification. The proposed technique is being compared with other computational intelligence dependent techniques of artificial neural network (i.e., multilayer perceptron-MLP). Results and different comparisons of proposed technique could work as an essential tool for fault diagnosis of WTGs.