Electrical arc surveillance and localization system based on advanced signal processing techniques

Abstract

This paper presents two methods applied in the context of surveilling and localizing electrical arcs which usually occur in photovoltaic power systems. The wavelet analysis has proved to be very powerful in transient signal detection, namely in multiscale edge detection, thereby we use this method to detect the electrical arcs and to localize them in time. Secondly, we applied the recurrence plot analysis (RPA) which presented succesful results in partial discharge time localization. In order to obtain the space localization of the electrical arc, we used the time-of-arrival of the arc at the acoustic sensor system. The system is composed of four sensors and the space localization is based on time difference of arrival, resulted from the time detection based on wavelet analysis and on RPA method.