Big Data Analysis Techniques for Power System Fault Detection and Diagnosis: A Review

Abstract

With the integration of renewable energy and the rapid development of smart grids, ensuring the reliability and security of modern power systems faces great challenges. This paper discusses how big data analytics and machine learning are transforming fault detection and diagnosis in dynamic power environments to address these complex issues. First, the main types and sources of fault data are introduced, and then the efficient data collection and preprocessing are emphasized. By adopting advanced feature engineering (including extraction, selection and dimensionality reduction), the diagnostic accuracy and system adaptability are significantly improved. Thru comparative analysis, the advantages and disadvantages of traditional machine learning, deep learning and hybrid algorithms in different fault scenarios are revealed. The paper then introduces the real-time implementation, focusing on the scalable data platform and how to combine it with the IoT-enabled smart grid. Finally, the paper also discusses important issues such as data quality, scalability, interpretability, and cyber-physical security. It also points out that the latest technologies such as edge artificial intelligence, federated learning, and digital twins represent directions with great potential for future development. These findings provide a comprehensive foundation for building a flexible, adaptive, and intelligent power grid management system, and provide valuable guidance for researchers and industry practitioners.