Tackling Renewable Energy Intermittency in Hydrogen-Based Solar Energy System Control Using DRL

Despite the potential significant demand for hydrogen produced from renewable energy, few studies have explored strategies for addressing intermittency challenges commonly experienced with renewable energy sources such as wind or solar generation. To address this gap, our study proposes a hybrid solar-battery-hydrogen system optimised to manage uncertainty using a combination of a machine learning technique, specifically deep reinforcement learning (RL), and linear programming. By leveraging this approach, we optimised hydrogen production using solar PV power while meeting hourly demand and adapting to disruptions in solar power generation. Our RL model trained with uncertainty in mind is designed to operate in various geographic locations and seasonal conditions, showcasing its versatility and potential for real-world implementation.