MACHINE LEARNING APPROACHES FOR FAULT DETECTION ANDDIAGNOSIS IN POWER SYSTEMS

Abstract

Power systems' dependability and effectiveness are crucial for an uninterrupted and safe supply of energy. Conventional techniques for power system problem diagnosis and detection often depend on expert knowledge and rule-based systems, which may not be sufficient to handle complex and dynamic circumstances. The use of machine learning (ML) techniques to improve problem identification and diagnosis in power systems is examined in this research. The article analyses numerous ML approaches, including supervised learning, unsupervised learning, and ensemble methods, applied to distinct datasets from power systems. These databases include data from sensors positioned strategically across the electrical grid, including voltage levels, current readings, and frequency fluctuations. A potential approach to improving the dependability and efficiency of vital infrastructure is the use of machine learning in power systems for issue identification and diagnosis. As the sector develops, it will be more important to take into account issues of interpretability, explainability, and ethical implications to guarantee the appropriate use of these technologies in the field of power system operations and maintenance