| Title |
Analysis of Distance-Dependent Feature Variations and Effective Feature Ranking in Residential Series Arc Fault Signals |
| Authors |
박종복(Jong-Bok Park) ; 지평식(Pyeong-Shik Ji) |
| DOI |
https://doi.org/10.5370/KIEE.2026.75.8.1971 |
| Keywords |
Residential Arc Fault Signals; Distance-Dependent Feature Variation; Effective Feature Ranking; Residential Loads; AC Series Arc |
| Abstract |
This paper presents an experimental analysis of distance-dependent variations in arc fault signal features and their rankings in residential electrical loads. In practical systems, current signals are measured at distribution panels rather than at the arc source, where characteristics can be affected by cable length and load-dependent propagation effects. To investigate this, arc and normal current signals were acquired at cable lengths from 0 m to 30 m. Time- and frequency-domain features were extracted and evaluated using the effect size measure η² to derive feature rankings across distances. The analysis focuses on a microwave oven and an electric fan. Results show that in the microwave oven, feature values and rankings vary significantly with distance, leading to reduced discriminative performance. In contrast, for the electric fan, key features such as variance, RMS, and normalized harmonics maintain consistent performance, showing high ranking stability across distances. These results indicate that distance effects depend on load characteristics and highlight the need for distance-aware feature analysis for reliable panel-level arc fault detection. |