DC Series Arc Detection Using FFT-Based Frequency Band Characteristics and Pattern Analysis

When a DC series arc fault occurs, it generates high-temperature plasma—often exceeding several thousand degrees Celsius—which can destroy insulation materials and cause electrical fires. To address this risk, this study proposes an arc fault detection method based on the frequency band energy characteristics of the current signal. The current signal is acquired at a sampling rate of 100 kS/s and transformed into the frequency domain using the Fast Fourier Transform (FFT). The 5–50 kHz range is divided into six sub-bands to extract relative frequency band energy ratios. These ratios are then subjected to pattern analysis by mapping the energy order across bands to all possible permutations (6! = 720). A sliding window approach is used to compute the temporal occurrence of dominant patterns. Experimental results demonstrate that changes in the distribution and frequency of these patterns between normal and arc states serve as effective indicators for DC series arc conditions. The proposed pattern analysis method enables precise arc fault detection, and it can serve as a robust feature extraction technique for machine learning models or be integrated into real-time arc monitoring systems in various electrical environments.