Multi-Feature Fusion Based Mechanical Fault Diagnosis for On-Load Tap Changers in Smart Grid with Electric Vehicles

With the proliferation of electric vehicles (EVs), the supporting facilities and infrastructure become new components in conjunction with conventional electrical appliances. These novel appliances, e.g., charging piles and energy storage devices, bring new features as well as challenges to the existing power grid. To enhance the accuracy of mechanical fault identification for on-load tap changers (OLTCs) in smart grid with EVs, a feature selection method for OLTC mechanical fault identification is proposed in this paper. This method relies on the multi-feature fusion and the joint application of the K-nearest neighbors algorithm (KNN) and the improved whale optimization algorithm (IWOA). By multi-feature fusion, the high-dimensional set of time-domain and frequencydomain characteristics as well as energy and composite multi-scale permutation entropy can be constructed. As a result, the maximum correlation minimum redundancy (mRMR) principle can be used to screen the sensitive feature subsets. Finally, IWOA is used to optimize the sensitive feature subsets, and KNN is used to classify the different types of optimal feature subsets. The experimental results show that the proposed method is at least 8% more accurate than the existing methods. The high-accuracy nature of the proposed method can accelerate the promotion of EVs and the establishment of intelligent transportation environments.