DETECTION, CLASSIFICATION, AND RATING OF ELECTRICAL FAULTS AND POWER QUALITY

Abstract

Power system faults are unavoidable and can seriously damage costly equipment such as generators, transformers, and motors. They may also cause overvoltages, high fault currents, power outages, explosions, and even loss of life. Therefore, an effective protection system is required for fast fault detection, classification, and localization. Accurate fault analysis helps ensure reliable power supply, reduce outages, and protect electrical infrastructure. This study presents a framework for detecting, classifying, and locating power system faults. Its objectives are to identify different fault conditions at various locations and resistances, determine the causes of interruptions, support quick power restoration, and reduce recurring failures. The study also improves understanding of protection system components to minimize equipment damage and service disruptions. Advanced machine learning techniques, including Convolutional Neural Networks (CNN), K-Nearest Neighbors (KNN), Random Forest (RF), and Decision Tree (DT), are used for fault analysis. These models identify fault types, locate faults, classify disturbances, and assess power quality across five classes. Implemented in MATLAB/Simulink on a test network, the proposed framework effectively addresses fault detection, classification, localization, and power quality assessment, improving the protection, stability, and efficiency of modern power systems.