Improving PEM Water Electrolysis Efficiency with ANN-Based Control to Handle Rapid Photovoltaic Power Fluctuations

Abstract

This research presents an innovative method for enhancing hydrogen production through proton exchange membrane (PEM) water electrolysis, powered by photovoltaic (PV) energy. The system is based on the Perturbation and Observation (P & O) method of Maximum Power Point Tracking (MPPT) with a boost converter to maximize energy capture, and a buck converter to stabilize DC voltage, ensuring compatibility with the proton exchange membrane electrolyzer. An artificial neural network (ANN)-based controller manages the buck converter, effectively minimizing the effects of solar irradiation fluctuations on electrolyzer performance. By using the adaptive learning capabilities of the ANN, the proposed approach increases the efficiency of hydrogen production under varying solar energy levels. Simulation results indicate that the ANN controller outperforms the conventional PI controller, reducing the mean absolute percentage error (MAPE) from 1.22 % to 0.95 %, decreasing overshoot from 12.84 % to 3.19 %, and achieving a faster settling time of 0.022 s compared to 0.023 s. This study advances renewable hydrogen production technologies, demonstrating that ANN-based control improves dynamic performance and contributes to the development of smarter, more resilient energy systems.