Fault Detection of AC Filter Breaker’s Closing Resistance based on Probabilistic Neural Network

The closing resistance fault in AC filter breaker will cause inrush current and affect stable operation of power system. To address this problem, this paper proposes a fault detection method for AC filter breaker’s closing resistance fault based on Probabilistic Neural Network. Firstly, this paper analyzes the causes and influencing factors of closing inrush current. According to the waveform’s characteristics, several fault features are chosen for detection. The fault feature dataset is formed by PSCAD/EMTDC. On the basis, Probabilistic Neural Network is applied to detect the closing resistance fault of AC filter breaker, where the neurons are determined by K-means clustering. Simulation results and actual record data verify the robustness and generalization ability of the proposed method-, which enhance its potential for practical application.