ABSTRACT This work aims to introduce an advanced adaptive protection scheme for modern power networks with renewable energy resources and various operational topologies. These topologies increase the difficulty of protecting the power grid as fault current levels changes, potentially causing delayed tripping, miscoordination or unnecessary outages. This study developed 34 network topologies using the IEEE 33‐bus system, incorporating a photovoltaic system (operated under two controllers), two wind turbines (operated under four types) and three battery energy storage systems. These network topologies are clustered using a new data‐driven approach that utilises the overcurrent relay (OCR) setting to deal with the operation variations of different 34 power grid topologies. The data‐driven clustering approach is compared to different clustering approaches, K‐means, K‐medoids and PSO‐based clustering. Principal component analysis and the water cycle algorithm are utilised to identify optimal relay settings using a non‐standard OCR characteristic that adapts to the actual fault level and topologies to achieve minimum tripping time. Overall, the proposed adaptive non‐standard relay scheme outperformed standard approaches and achieved lower tripping times, with reductions of 10% to 25%.
Shihab et al. (Thu,) studied this question.