Advanced signal processing techniques for detection and localization of electrical arcs

This paper presents several methods applied for the detection and localization of electrical arcs measured while survelling photovoltaic power systems. Firstly, we proposed the use of energy detectors for transients (spectrogram and wavelet) and compared them with statistical methods (Maximum Likelihood Estimation (MLE)), classical signal processing methods (Matched Filter, Zero Crossing), but not lastly with a more recent method, Recurrence Plot Analysis (RPA), which has already proved its efficiency. Afterward, we studied the precision of these methods in the localization problem. We used a four sensor detector and estimated the position of the electrical arc based on the time of arrival (TOA) obtained from the each technique.