A comprehensive review of islanding detection methods: Transition to intelligent classifier based approaches

When a distributed energy resource (DER), such as a grid-tied inverter, identifies an island that is isolated from the main and stops energizing the power system, this is known as unintentional islanding. Maintaining local load supply is a significant obstacle in contemporary distribution networks. When the local generation is aligned with the local load, requiring islanding detection, it becomes more challenging, creating a non-detection zone. Therefore, stable, reliable, and effective islanding detection systems are crucial to prevent islanding and ensure the electrical grid runs safely. This study reviews the progression of islanding detection approaches, grouping them into remote, local, and intelligent classifier-based approaches. While traditional schemes have dominated historically, intelligent classifier-based approaches are gaining prominence due to their enhanced capabilities. This paper discusses the move toward intelligent approaches, highlighting their key advantages, difficulties, and future avenues for research. This paper is comprehensively analysing- AI based islanding detection techniques developed over the last decade with a strong emphasis on feature selection, critical electrical parameters, existing challenges, and emerging future research needs. Finally, an Optuna-optimized SVM case study is included to illustrate the detection performance of intelligent IDSs in this review.