Series DC Arc Fault Detection Using Machine Learning Algorithms

The wide variety of arc faults induced by different load types renders residential series fault detection complicated and challenging. Series dc could cause fire accidents adversely affect power systems if not promptly detected. However, in practical systems, they are difficult to detect because a low current, absence zero-crossing period, various abnormal behavior based on loads controllers. In particular, conventional protection fuses may be activated when occur. Undetected false operation potentially lead damage property human casualties. Therefore, it is imperative develop system for DC the reliable efficient such systems. this study, several typical loads, especially nonlinear complex as electronic were chosen analyzed, five time-domain parameters current—average value, median variance RMS distance maximum minimum values—were detection. Various machine learning algorithms used their accuracies compared.