Machine Learning Based Hydrogen Electrolyzer Control Strategy for Solar Power Output and Battery State of Charge Regulation

Photovoltaic (PV) power is an extensively used renewable energy resource, but its intermittent nature affects the power supply quality as it results in issues such as frequency aberrations and voltage variations. Battery Energy Storage Systems (BESS) are utilized to smooth out and resolve the fluctuation issues. However, a control method is required for BESS charging level regulation to prevent the need for larger storage systems and to extend its operational life through controlled charging/discharging. This paper proposes a novel solar power and battery state of charge (SoC) control technique through the incorporation of hydrogen electrolyzer (HE) fuel cell system and BESS. A machine learning based controller (MLC) is designed for dynamic control of the HE output to allow the dispatching of firmed PV power with controlled battery charging/discharging. The MLC takes the fluctuating power and various battery parameters as inputs and intelligently controls the HE output while obeying the imposed constraints. Results conclude that the MLC greatly reduces the PV power fluctuations and a comparison between the fuzzy logic control for SoC regulation shows that the MLC has better SoC management capability. The proposed methodology promotes the integration of hydrogen into the energy mix as a means for providing controlled solar power.