2nd China International Youth Conference on Electrical Engineering (CIYCEE 2021)

17 / 2021-09-17 09:27:57

Latent fault detection method for medium voltage cables based on Kizilcay arc model and KNN

Latent cable fault; Kizilcay arc model; Phase space reconstruction; Fractal theory; KNN algorithm classifier

Power system and automation > Power system fault simulation and risk assessment

Cable plays an important role in urban medium voltage distribution networks. Detecting and locating latent cable faults is of great significance for preventing permanent cable faults, reducing power outages caused by faults, and ensuring the safe operation of power systems. In this paper, based on Kizilcay arc model, an equivalent model suitable for cable latent fault in small resistance grounded distribution system is established, and a typical 10kV small resistance cable distribution network model is built in PSCAD / EMTDC platform to compare and analyze the effects of different fault parameters on the characteristics of latent fault arc. Secondly, based on the phase space reconstruction theory and fractal theory, the latent fault characteristics of medium voltage distribution cables are extracted. Based on latent fault feature vector and KNN algorithm classifier, a new latent fault detection method for medium voltage distribution cable is proposed. The effectiveness of the proposed latent fault detection method is verified by PSCAD / EMTDC simulation.