Real Defect Partial Discharge Identification Method for Power Cables Joints Based on Integrated PJS-M and GA-SVM Algorithm with Multi-source Fusion

Lingxuan Zhang^*Yiyang Zhou沈炯 姚Jialuo ChaiYingjing ChenZhousheng Zhang^*

• Shanghai University of Electric Power, Shanghai, China

The majority of previous studies on the defects of 10 kV cable intermediate joints have focused on typical defects and relied on single sensors to collect feature quantities, resulting in incomplete characterization of defect features, and consequently, lower defect type recognition rates. Based on the types of real defects that may occur during the actual manufacturing process of cable intermediate joints, this study established three real-type partial discharge (PD) models. A combination of High-Frequency Current Transformer (HFCT) and Ultra High Frequency (UHF) sensors was used to acquire time-domain waveforms, frequency-domain spectra, and Phase-Resolved Partial Discharge (PRPD) patterns from the models, from which feature quantities were extracted. By incorporating an adaptive PJS-M weighting coefficient and a correlation-analysis-based dynamic correction mechanism into the conventional GA-SVM framework, a Genetic Algorithm Weighted Support Vector Machine (GAW-SVM) model was developed. The recognition performance of the proposed model was subsequently compared with several mainstream SVM optimization algorithms reported in recent years. Experimental results show that, under multi-source feature fusion, the GAW-SVM achieved a defect-recognition accuracy of 98.84%, representing improvements of 3.49%, 2.33%, and 1.17% over the GA-SVM, PCA-SVM, and PSO-SVM algorithms, respectively. These findings demonstrate that the proposed method enables high-precision identification of complex real-type defects under multi-source feature conditions, providing a reliable diagnostic basis and technical reference for partial discharge detection in cable intermediate joints.