Hardware Implementation of Time-Frequency Analysis Based DC Arc Fault Detection Algorithm in Photovoltaic Systems with Different Power Electronic Equipment

With the increase of system capacities and voltage levels in photovoltaic systems, a large amount of power electronic equipment has been gradually applied, which introduces much complex electromagnetic interference. Switching frequencies and control strategies of different power electronic equipment would interfere with the arc fault detection, which eventually make the detection algorithm invalid. Especially for new GaN-based converters with the high switching frequency, arc fault characteristics have not been studied. In this paper, arc fault current waveforms for different types of inverters and converters are obtained by their own current sensors. From the perspective of the time-frequency domain, the interference of different types of power electronic equipment on the arc fault detection is analyzed. Thus, the common arc fault detection variable is extracted. Based on the machine learning method, the arc fault detection algorithm could be proposed with high detection accuracy and wide application range. Finally, the arc fault detection algorithm is integrated into the control system of power electronic equipment. Hardware test results show that the arc fault could be extinguished in 1s, and the detection accuracy is higher than 95%.