Low-Voltage AC Series Arc Fault Identification Method Based on the Randomness of Differential Voltage at Double-Ended Monitoring Points

In low-voltage ac distribution systems, differences in load categories have a significant effect on the waveform of ac series arc currents. In contrast, arc voltage characteristics are less affected by load type, so it is easier to establish a uniform fault detection criterion by identifying arc voltage. Based on analyzing the characteristics of arc voltage and differential voltage at upstream and downstream monitoring points, a fault detection algorithm based on the randomness of differential voltage at double-ended monitoring points is proposed. The method can effectively identify the line voltage drop and fault arc voltage and exclude the impact of load fluctuations. In this process, while the differential voltage at the monitoring points is being denoised, the high-frequency fault characteristics in the fault signal are also suppressed to a certain extent. A band-stop filter with zero as the center is, therefore, proposed to amplify the fault characteristics in the high-frequency range. Ultimately, the arc eigenvalues are obtained, and the detection strategy is determined by measuring the similarity and difference of the filtered signals between cycles. The experimental results show that the method can effectively achieve series arc fault detection in low-voltage lines. It also has good recognition accuracy and generalization ability for unknown loads with strong noise immunity. It is a new detection method in electric power Internet of Things (IoT) technology with high engineering practical value.