A Comprehensive overview of Classification-enabled Machine Learning Algorithms for Islanding Detection Techniques

Abstract In modern distribution networks, addressing the issue of unintentional islanding – characterized as the inadvertent disconnection of distributed generation sources from the utility grid – continues to present a significant challenge. This phenomenon raises concerns that warrant further investigation due to its implications for system reliability and operational safety. The identification of islanding events is particularly complicated when local generation closely aligns with local demand, making detection difficult. The development of precise, rapid, and dependable methodologies for the detection of islanding in renewable and distributed generation systems requires compliance with rigorous standards. The current body of literature delineates an array of strategies for islanding detection, which can be systematically categorized into three primary approaches: (i) remote detection techniques, (ii) local detection methodologies, and (iii) machine learning-based classification-enabled intelligent classifiers. Recent advancements have garnered significant attention regarding the enhanced characteristics and benefits of intelligent methodologies in contrast to traditional approaches. This research provides a comprehensive overview of the transition from traditional techniques to intelligent islanding detection methodologies. Moreover, it elucidates the primary challenges, benefits, limitations, and prospective directions for research in intelligent detection schemes. Furthermore, this study provides a comprehensive and impartial analysis of intelligent classifier-based strategies for islanding detection that have been developed over the last decade. This research further examines various feature selection techniques and identifies the parameters most employed for efficient islanding detection. In conclusion, this comprehensive study presents a discussion of the findings obtained, along with strategic recommendations for future research initiatives within this field.