Optimization of a Hybrid Power System Employing Solar and Load Predictions

This paper investigates the benefits of model prediction in a hybrid power system, which comprises solar cells, and a proton exchange membrane fuel cell (PEMFC). The PEMFC is a backup power, providing auxiliary power when necessary. We applied experimental data to build a MATLAB Simscape Electrical $^{TM}$ model that could simulate system responses, where we found hydrogen consumption might be reduced by foreseeing the solar and load data. Therefore, we applied machine learning techniques to develop two prediction models that can foretell solar radiation and load responses with an accuracy of 96.34% and 93.35%, respectively. We then integrated the prediction models with the hybrid power system. The results showed that model predictions could prevent unnecessary hydrogen consumption and reduce system costs by 40.38% compared to the original system configuration. We also designed an experiment to show the feasibility of integrating these prediction models with the hybrid power system. The results demonstrated the benefits of model prediction for the hybrid power system.