Intermittent series direct current arc fault detection in direct current more‐electric engine power systems based on wavelet energy spectra and artificial neural network

The move towards More-Electric Aircraft (MEA) and Engine (MEE) systems has resulted in system integrators exploring the use of Direct Current (DC) for primary power distribution to both reduce energy conversion stages enable parallelling non-synchronised engine off-take generation. A prevalent challenge utilising DC is safe management arc faults. Arcing imposes a significant threat aircraft can result critical damage. Series intermittent faults are particularly challenging detect due lack zero current crossing coupled with intermittency caused by in-flight vibrations. Additionally, several state-of-the-art fault detection methodologies fail address handling system-specific conditions such as dynamic loading stability requirements. This paper addresses these unique MEA/MEE issues through proposal new generalised real-time methodology (WaSp) based on time–frequency domain-extracted features applied feed-forward Artificial Neural Network (ANN). outlines analysis domain features. Simulation-based case studies emulating range on-board EPS presented show proposed potential highly accurate performance fast times.