DWT-Based Detection and Characteristic Analysis of Partial Disconnection Faults Under Non-linear Load Condition

Series arc faults from partial disconnections are a major cause of electrical fires, yet extremely difficult to detect as their currents resemble normal loads. Under non-linear loads like Switched-Mode Power Supplies (SMPS), massive switching pulses mask high-frequency arc signals, causing critical Arc Fault Circuit Interrupter (AFCI) nuisance tripping. This paper proposes a Discrete Wavelet Transform (DWT)-based multi-resolution analysis to accurately detect series arcs under these complex conditions. To secure highly realistic simulation data, a 'Stochastic Modified Cassie Model' reflecting the physical intermittency of arc plasma was established. While conventional Fast Fourier Transform (FFT) failed due to SMPS harmonic masking and a lack of time-localization, the proposed DWT method successfully filtered the 15A-class SMPS pulses and 60Hz fundamental wave into the approximation coefficients. Consequently, it precisely isolated the asymmetrical, random high-frequency arc signals exclusively within the detail coefficient bands. This study mathematically proves that the DWT algorithm reliably diagnoses series arcs in non-linear environments, providing fundamental data for designing future intelligent fire prevention systems and mitigating AFCI nuisance tripping.