Optimizing Energy Management Systems for Smart Grids With Hybrid Solar and Hydrogen Fuel Cells Using IGANN–HSWO Approach

ABSTRACT An ideally designed hybrid renewable energy system (HRES) that is not dependent on the grid, incorporating photovoltaic (PV), hydrogen fuel cells (FCs), and battery storage integrated with the utility grid. Because of demand fluctuations, expensive component costs, and the unpredictable nature of resources, designing an ideal hybrid energy system is challenging. This manuscript gives an integrated energy management framework for a grid‐connected hybrid energy system comprising PV, hydrogen FC, and battery storage. The proposed approach combines an interpretable generalized additive neural network (IGANN) for load forecasting with a hybrid swarm whale optimization (HSWO) algorithm for optimal energy dispatch. The IGANN model captures nonlinear load patterns while maintaining interpretability, enabling a better understanding of the influence of input variables such as solar irradiance and demand variations. The HSWO algorithm utilizes the predicted load to perform adaptive optimization under system constraints, including power balance and storage limitations. Simulation outcomes establish that the proposed technique achieves comparable annual cost performance relative to existing approaches while providing improved adaptability and transparency in decision‐making. The integration of interpretable forecasting with constraint‐aware optimization enhances the reliability and flexibility of smart grid energy management systems. “In addition, interpretability of the IGANN model is validated using shape function and feature importance analysis.” The proposed method's performance is then tested utilizing the MATLAB platform, where it is compared with other existing approaches, such as artificial bee colony–particle swarm optimization (ABC–PSO), fuzzy logic control (FLC), and deep neural networks (DNNs), to evaluate its effectiveness and efficiency. The proposed method cost is $14 458, the ABC–PSO method cost is $15 514, the DNN method cost is $14 771 and the FLC method cost is $14 466. From the result, the proposed method shows lower annual cost compared to the existing approaches.