Framework for Classifying Fault Types in Power Networks Using Machine Learning

Abstract

The reliable operation of power systems hinges on the swift and accurate classification of fault types. Conventional fault type classification methods typically employ intricate rule-based or model-based systems, which may struggle to adapt to the dynamic and diverse nature of fault scenarios. In recent years, machine learning (ML) algorithms have emerged as promising tools for enhancing fault type classification processes in power systems. These algorithms excel in learning patterns from data without the need for explicit programming. This study provides a comprehensive review and analysis of various ML techniques applied to fault type classification in power systems. It addresses key challenges such as the variability of fault types, the non-stationary behavior of signals, and the requirement for real-time response. The paper explores a range of machine learning algorithms including support vector machines (SVMs), neural networks, decision trees, and ensemble methods, highlighting their respective strengths and weaknesses in fault type classification scenarios. Moreover, the study offers insights into critical aspects like feature selection and pre-processing techniques that play pivotal roles in optimizing the performance of ML models in fault detection tasks. Evaluation metrics such as accuracy, sensitivity, specificity, and computational efficiency are used to assess the efficacy of each approach. © 2025 IEEE.