Research on AC Arc Fault Detection Method Based on Wigner-Hough Transform and NG-Boost

Research shows that arc fault is one of the extremely common causes of electric fire. Due to the characteristics of series arc fault, the traditional overload, short circuit and leakage protection devices in low-voltage distribution system cannot precisely detect the arc fault in the circuit. At the same time, the existing low voltage AC arc fault detection is easily affected by the nonlinear load in the system, resulting in misoperation or rejection. To solve the existing problems, an AC series arc fault feature extraction method which based on Wigner-Hough transform is proposed in this paper. The WHT peak change ratios of current in adjacent periods are selected as the features, and the natural gradient lifting algorithm (NG-Boost) is used for state judgment. The off-line test results show that the proposed method can realize the fault detection under 7 load conditions.