Partial Discharge Feature Extraction of 10kV XLPE Cable Using EEMD

Abstract

Partial discharge (PD) is damage to the XLPE Cable, to study the PD of 10kV XLPE Cable, this paper mainly studied the three typical insulation defects of power cables, which were metal impurities, main insulation scratches and main insulation water inflow. The feature extraction of PD signal was carried out by the method of Ensemble Empirical Mode Decomposition (EEMD). The amplitude and energy of the intrinsic mode functions (IMF) component obtained from EEMD decomposition of the three kinds of cable typical insulation defects were studied, and their similarities and differences were analyzed and compared. The concept of energy entropy was used to extract five kinds of characteristic parameters as the feature quantity. The effectiveness of feature extraction was verified by Convolutional Neural Networks (CNN) and other algorithms.