Evaluation of Feature Extraction Technique for DC Arcing Fault at Aerospace Power Management and Distribution (PMAD) System

The arcing faults on cables of aerospace power systems represent a major safety concern for spacecraft. To ensure the normal operation of the Power Management and Distribution (PMAD) system, it is necessary that remedial control strategies are developed and implemented. Most of the research in this area deal mainly with arcing faults in terrestrial distribution power systems that operate as AC systems rather than DC systems. The paper investigates the application of a new digital signal processing technique – Fast Fourier Transform (FFT) based energy spectrum estimation – to analyze the recorded signals of DC arcing faults. The data is collected from the DC arcing experiment at the NASA Glenn research center. The voltage level range is from 50Vdc to 150Vdc. This paper presents a feature extraction approach to transform the signal representation from time-domain into frequency-domain. Results demonstrate the distinctive difference between the pre-fault signal and post-fault signal. The proposed feature extraction algorithm can be integrated into fault detection algorithms for fault detection.