Application of feature extraction method based on 2D — LPEWT in cable partial discharge analysis

Partial discharge detection is important for identifying insulation defects of power cables, and the extraction of effective characteristic parameters is the emphasis of the study. In this paper, a feature extraction method based on 2 Dimensions Littlewood - Pale empirical wavelet transform (2D-LPEWT) is proposed to realize the accurate identification of different types of defects in cable discharge. By constructing a cable-insulated partial discharge detection platform, three-dimensional spectrum of four typical defect models was decomposed by 2D-LPEWT. And then Tamura features, moments and entropy characteristics are extracted from the obtained wavelet coefficients sub-graphs, and the effects of different feature extraction methods on performance of KNN, decision tree and SVM are discussed. Results show that the proposed feature extraction method can achieve high recognition accuracy under different classifiers, and has good practicability.