Research on live detection technology of distribution network cable insulation deterioration state based on harmonic components

Ran Hu,Haisong Xu,Xu Lu,Anzhe Wang,Zhifeng Xu,Yuli Wang,Daning Zhang

DOI: https://doi.org/10.1049/gtd2.13238

2024-08-18

IET Generation Transmission & Distribution

Abstract:This study enhances the accurate diagnosis of cable insulation conditions by real‐time monitoring harmonic currents in distribution network cables within urban power grids, serving as an effective complement to traditional offline diagnostic methods. Through the establishment of an experimental platform and electromagnetic simulation model, the magnetic flux evolution patterns under typical defects were investigated. A method for assessing the degradation of cable insulation was developed based on harmonic feature data. Defect type identification for moisture‐affected and aged cables was achieved through expectation‐maximization clustering analysis, providing means for the effective detection of high‐risk cables. Due to the limitations imposed by urban power grid outages for maintenance, on‐line harmonic current detection technology for distribution network cables is expected to become an effective supplement to traditional offline diagnostic methods, enhancing the real‐time diagnosis of distribution network cable insulation conditions. This study established a 10 kV distribution network cable test platform and prepared typical defective cables subjected to moisture and long‐term thermal aging. Using COMSOL finite element electromagnetic simulation, the magnetic flux evolution laws of the cable insulation under typical defects were obtained. Experimental tests provided the harmonic current characteristics and statistical features of cables with typical defects. Based on these data, a method for analysing the degradation degree of distribution network cables was constructed using least absolute shrinkage and selection operator (LASSO) regression analysis. Furthermore, a defect‐type identification method based on cluster analysis was proposed. Results indicate that the odd harmonics and the 4th harmonic of the distribution network cable's harmonic current are closely related to the cable's degradation state. A model integrating principal component analysis (PCA) data dimensionality reduction and expectation‐maximization clustering analysis achieved a recognition accuracy of up to 75.64% in distinguishing between moisture‐affected and normal cable states. The proposed on‐line detection and evaluation methods can effectively identify high‐risk cables with latent defects.

engineering, electrical & electronic