Energy Management Solution for Islanding Based on a Dynamic Neuro-Fuzzy-Optical Microscope Algorithm

Abstract

Ensuring continuity of service is a primary objective in power systems. In grid-connected microgrids (MGs), islanding poses a significant threat to this continuity. Conventional approaches mitigate islanding by disconnecting the MG immediately after separation from the main grid to prevent overload and ensure safety, but this results in service interruption. This study proposes a dynamic islanding management strategy that maintains uninterrupted service using an optimal dynamic neuro-fuzzy–optical microscope algorithm (OMA). The method integrates a convolutional neural network (CNN), fuzzy logic (FL), and the novel OMA optimizer in a two-stage framework. In the first stage, the CNN detects islanding based on active current and voltage measurements at the point of common coupling (PCC) and their dominant harmonic components, obtained from a hybrid MG model. This model is comprising solar panels, wind turbines, a biomass generator, and a storage system. Signal and image processing techniques prepare the measurements for CNN implementation. Upon islanding detection, the second stage is activated, where FL predicts the penalty factor and OMA optimally manages economic power sharing between the grid and the MG. This integration enables safe load coverage without damaging MG components. Performance benchmarking against Quadratic Interpolation Optimization (QIO) and Hunger Games Optimizer (HGO) demonstrates that OMA achieves higher accuracy, faster convergence, and lower execution time. Validation across five scenarios under normal, islanding, and risky operating conditions confirms the method’s effectiveness, reliability, and economic benefits, achieving a 223.7% revenue improvement over the baseline with the shortest execution time. The proposed approach offers a robust and intelligent solution to the islanding problem, ensuring continuous and cost-effective microgrid operation.