A Deep Learning-Based Islanding Detection Approach by Considering the Load Demand of DGs Under Different Grid Conditions

Abstract

Islanding detection is a very important issue in the integration of renewable energy systems with the grid. In recent years, especially artificial intelligence and deep learning-based islanding detection methods have come to the fore in terms of providing reliable power quality. In this study, a deep learning-based islanding detection approach by considering power quality and load demand problems is proposed. It is aimed to effectively detect the islanding condition which occurs as a result of unintentional disconnection of distributed generation (DG) systems from the grid. In the proposed approach, a deep learning-based islanding detection method is developed, taking into account the faults and power quality events occurring on the load side like considering asynchronous motor startup, capacitor switching, etc., conditions that are not possible to easily detect by conventional islanding detection methods. With the developed method, it is seen that the islanding event can be distinguished from the power quality events that occur on the grid, even under noisy signals. In this way, the power quality of the grid is increased and the performance of the DG in dynamic load behavior is developed. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.