Arcing Fault Type Identification with Light Spectrum

Abstract

Currently, most modern Arc Flash relays use light detection to detect an arcing event, optionally they add current detection for security. Traditional scheme relies on the current for fault classification. This paper proposes an approach to identify arcing fault types by using the light spectrum emission from the materials or coating of the materials during the incident. This is based upon the distinct spectra that is emitted from excited material that can be used as signatures to identify the fault type. As one of the characteristics of arcing faults, intensive lights would be produced. Therefore, using the different type of conductors and coatings on different phases has the potential to be used to identify arcing fault types by analyzing the light spectrum during the event. In this study, the conductors made of copper and aluminum and the copper conductor with coating Tin are used as the example to test the effectiveness of the proposed approach. Light spectrum of each case during the arc flash will be measured and recorded by a spectrometer. By using the light spectrum of arcing fault conductors, types of arcing fault could be identified by using General Regression Neural Network (GRNN). The application of this study could improve the speed to identify the fault type and reduce the downtime.