DC Arc Failure Detection based on Division of Time and Frequency Components using Intelligence Models

Abstract

This study investigates an approach related detection of series arc faults in the DC lines through the utilization of features extracted from the difference between odd and even components of the signal, integrated with intelligence models in diverse domains. Series DC arc faults pose significant safety risks in various electrical systems, necessitating robust detection methods. In this research, the authors propose a novel approach that leverages the unique characteristics of the signal’s odd and even components to enhance fault detection accuracy. The methodology involves preprocessing the signal to extract relevant features capturing the discrepancy between odd and even components, which are then used as inputs for AI models. These models are trained to classify fault and non-fault conditions based on the extracted features. The integration of feature extraction from odd and even signal components with AI models offers a promising solution for heightening the reliability and efficiency of DC arc error recognition systems in various industrial and residential applications.