Abstract: Electrical fires caused by arc faults are a frequent occurrence. This paper presents a comparative analysis of detection algorithms of Arc Fault Detection Devices (AFDDs) available on the market. The typical household electrical characteristic loads are selected, and an evaluation system for AFDD performance is established. Three key performance indicators—coefficient of variation of breaking time, breaking time speed, and the number of false trippings—are introduced and quantified. The entropy method is applied for weight allocation to develop a performance-based quantitative assessment method. The evaluation is validated with AFDD based on Fourier transform detection method, wavelet transform detection method and artificial neural network detection algorithm. The testing and analysis are conducted in accordance with the national standard GB/T 31143-2014 as well as the performance quantification evaluation method proposed in this paper. The verification results have identified AFDDs that fail under specific residential characteristic loads, demonstrating that the performance quantification evaluation system proposed in this paper can effectively measure the performance of AFDDs under typical residential electrical characteristic loads, offering valuable insights for real-world electricity usage scenarios in households.

Key words: arc fault detection devices, performance quantification assessment system, electrical fire