Deep learning-based intelligent islanding detection for grid-connected photovoltaic systems using convolutional neural networks

Abstract

The increasing integration of photovoltaic (PV) systems into smart grids requires fast and reliable islanding detection to maintain grid stability and operational safety. Conventional detection methods often face challenges such as delayed response, reduced accuracy, and large non-detection zones under varying operating conditions. This paper proposes an intelligent islanding detection method for grid-connected PV systems using advanced artificial intelligence and deep learning techniques. Electrical parameters including voltage, current, frequency, and power signals are analyzed using signal processing methods and classified through a convolutional neural network (CNN) model developed in MATLAB/Simulink. Simulation results demonstrate that the proposed AI-based approach achieves rapid and accurate detection of islanding events with improved sensitivity and reduced false detections compared to conventional techniques. The proposed framework enhances the reliability, safety, and protection performance of modern photovoltaic power systems integrated with smart grids.