Research on Intelligent Diagnosis System for High-voltage Cables Based on Partial Discharge Characteristics

High-voltage cable insulation state directly determines the service life of these cables, and after the periods of producing, laying and long-term operating, there is likely to introduce impurities, air bubbles into the insulation layer. These defects will bring partial discharge, which will damage insulation strength, and cause cable failures. Due to the partial discharge test can achieve the purposes of early cable insulation state assessment, cable defects detection and operational reliability guarantee, lots of cable partial discharge detection equipment have emerged at home and abroad. However, the equipment of different manufacturers have their own characteristics, so the use of these equipment also bring problems such as high requirements for technicians, and low data utilization. To solve these problems, it becomes urgent to build a unified diagnostic platform, upgrade the identification algorithm, maintain a partial discharge defect library, and access the online monitoring data of partial discharge. Based on these needs, this paper designed and implemented a partial discharge data diagnosis system by applying convolutional neural networks for massive field measured partial discharge sample data. The training and verification of these data were completed, and the model which proved has a high recognition rate, good theoretical and practical significance through simulation experiment for partial discharge diagnosis was output.