Identification of Series DC Arc Fault Location in MPPT Photovoltaic System Based on Fast Fourier Transform and Artificial Neural Network

The growth of Solar Energy is being increased and developed. One of them, the Maximum Power Point Tracking (MPPT) is mainly used to get a maximum output power in unstable weather conditions. However, a series of DC arc fault causes damaged insulation and fire. This paper will discuss the identification of series DC arc fault location in conventional MPPT photovoltaic array based on PSIM and MATLAB simulation. Series DC arc fault will be analyzed using a current signal parallel capacitor on a photovoltaic array. Furthermore, those signals will be processed with the Fast Fourier Transform and Artificial Neural Network to know where the series DC arc fault occurs. The proposed location identification method gives a 98.2% accuracy rate when implemented in a conventional MPPT photovoltaic system. The average sum of frequency DC-link capacitor current ratio between series DC arc fault occurs for all proposed locations against normal's condition is 1.39807:1. 18 different characteristic testing data validated the simulation result, and it shows the proposed method is more accurate from a comparison between possible neural network training methods.