Abstract

Abstract—Arc fault is one of the most important reasons of electrical fires. The common characteristics of series arc faults are always submerged in load currents which are acquired as mixed signals. It is difficult to extract useful arc fault characteristics through conventional approaches. Hence, novel approaches based on blind source separation and harmonic analysis are developed for characteristic research of series arc faults. At first, a low-voltage arc fault experimental platform was set up to generate arc faults of different loads, and a large number of load currents were acquired to analyze fault characteristics. When series arc faults occurred in circuits, currents would distort a little. Then, in order to obtain current minor changes, blind source separation (BSS) was used to separate distortions from observed mixed signals. Finally, common characteristics of series arc faults were found. The amplitudes of odd harmonic components in arc fault currents change rapidly. The total odd harmonic ratio fluctuates tempestuously with inconstancy but increases a lot. These harmonic characteristics provide very important bases for arc fault detection.