Multi-Criteria Analysis of Series Arc Faults in a Low-Voltage Electrical Network

In this paper we present a series arc fault detection method in a low voltage DC electrical network (270 Volt DC). The detection is based on a multi criteria analysis on the line current. Analyses are applied to the current or residual signal in order to study the fluctuations produced by the initiation and subsequent maintenance of the arc. First, the frequency analysis is performed on the total line current. Extraction of the current baseline allows arc noise to be visualized. Next the analysis is performed on current segments using the sliding variance, the autocorrelation of the signal and the evaluation of the power of the residual. The detection method is tested mainly on resistive loads (from 3A to 24 A). In this article, the arc faults are produced by the separating electrodes procedure. Analysis of the different approaches show that the series arcs faults can be successfully detected.