MACHINE LEARNING BASED FAULT LEVEL ANALYSIS FOR DISTRIBUTION SUBSTATIONS

Abstract

Fault level analysis in the distribution substations is an important and key part of the power system's protection and reliability. Disturbances like three-phase, line-to-line, and line-to-earth faults can drastically affect system performance, damage the equipment, and cause extended interruptions. An accurate fault identification is needed to select appropriate protection devices and ensure prompt fault fixing to minimize downtime. Traditional analysis methods heavily rely on electrical parameters, which may not be available in many functioning environments. This paper presents a machine-learning-based approach for predicting and categorising the fault types in distribution substation studies using the available parameters in hand. The proposed method helps power companies find faults more easily, make their systems stronger, and take steps to fix problems before they happen.