ABSTRACT As one of the important links in energy transition and global energy interconnection, electric vehicles (EVs) are becoming increasingly major research topics in the field of energy. Partial discharge (PD) identification under repetitive impulsive voltages plays a pivotal role in evaluating the health of electric vehicle (EV) motor insulation. However, the extraction process mainly treats PD signals as independent events, whereas the temporal correlation of their typical time series remains unexplored. Therefore, this study employs the Toeplitz inverse covariance‐based clustering (TICC) algorithm, which accounts for the inherent temporal correlation in time‐series data. Based on temporal features, the PD signals under nanosecond pulse voltages are segmented and extracted into four characteristic phases: pre‐discharge, commutation noise, partial discharge and post‐discharge. The results demonstrate that under optimised parameter settings (sliding window w = 15, number of clusters K = 4, and regularisation parameter λ = 0.02), the proposed clustering model achieved effective clustering of PD signals at optimal segmentation points. The model has good robustness and interpretability for PD signal identification under pulse voltages with different polarity and amplitude. This study provides both theoretical insights and practical guidance for accurate discharge signal extraction and motor insulation system design. Furthermore, it offers technical support for insulation condition monitoring in low‐altitude aircraft.
Wang et al. (Mon,) studied this question.