Series arc fault detection based on current fluctuation and zero-current features

Abstract

A series arc fault detection method based on current fluctuation features and zero-current feature fusion is proposed. In view of the problems of series arc faults, such as the strong concealment and high randomness, relatively small current amplitude easily annihilated by the load current when it occurs, and influence of the load properties, a low-voltage single-phase alternating-current (AC) series arc fault experiment platform is developed with reference to the UL1699 standard. The four-cycle current of the electric circuit is collected and the zero-current time proportional coefficient and normalized mean square error coefficient are calculated. A fuzzy logic device is then used to fuse the two coefficients to obtain the comprehensive feature identification coefficient of the series arc faults. The zero-current time proportional coefficient is combined and compared to the empirical threshold to determine whether there are series arc faults. This method has a high identification rate of 100% for series arc faults, without misjudgment or missed judgment when the GB14287.4 standard recommended load is used in the low-voltage single-phase AC power circuit.