DC arc fault detection method based on K-line and spectrum integration difference of current

DC fault arc occurring in low voltage DC systems such as photovoltaic (PV) system and battery energy storage systems is difficult to be extinguished. The traditional arc fault identification approaches have low recognition accuracy for the series arc. Therefore, a reliable approach is needed to detect DC fault arc timely and accurately. In this paper, the time and frequency domain characteristics of the current signal of a low-voltage DC system circuit are analyzed. A comprehensive algorithm based on the K-line and spectrum integration difference of the current is proposed, which can detect the low voltage DC fault arc, and effectively distinguish four states of the circuit: normal current, DC arc, switch action and load mutation. The real time arc detection test is performed using a microcontroller STM32F407. In the experiment, the fault arc can be detected and isolated within 37 ms, which is in line with the UL1699B standard, and the identification accuracy can reach 99.17%.