DC Series Arc Fault Detection Using Fractal Theory

Abstract

Arc faults, often caused by insulation or component failure, result in a discharge of electricity through the air between conductors. These failures are often the cause of electrical fires and pose an enhanced risk to system reliability, and this is becoming a growing problem with the uptake of more electric automotive and aircraft technologies. DC series arcs are of a particular concern as they do not trip existing circuit overcurrent protection. Arc detection is becoming increasingly difficult as DC voltages increase to meet the higher power demands of renewables, transport and series applications. This paper proposes a novel method to detect DC series arcs by monitoring the fractal dimension of the supply and load current and voltage waveforms. DC series arc faults were reproduced across a range of different setups using a 42V supply and a resistive-inductive load. The Windowed Fractal Dimension (WFD) method; implemented in MATLAB, shows a clear change in fractal dimension when an arc is sustained, providing both a means of arc fault detection and evidence that arcs have fractal properties.