Hybrid wind-hydrogen plants have multiple revenue sources, subject to uncertainties and trade-offs. As a consequence, maximizing their overall profitability is a challenging optimization problem. Since electricity is typically traded in advance of its actual power generation, weather forecasts play a crucial role in the power trading strategy. Additionally, the trading and control strategies of other market players influence the imbalance of the public grid and, as a result, have an impact on the revenue that can be generated by grid balancing. To address this challenge, we propose a novel approach based on multi-agent reinforcement learning (RL), with two RL agents: one for day-ahead power trading, and a second for real-time power control of the wind farm and the electrolyzer. The RL system is trained using real-world data from a large offshore wind farm in the Belgian North Sea. Results demonstrate the effectiveness of this approach, with the RL agents collaboratively maximizing the total operational profit of the hybrid plant, achieving a significantly higher profitability compared to conventional methods.
Ally, S, Verstraeten, T, Nowé, A & Helsen, J 2025, 'Optimal day-ahead trading and power control for a hybrid wind-hydrogen plant with multi-agent reinforcement learning', Journal of Physics: Conference Series, vol. 3025, no. 1, 012022. https://doi.org/10.1088/1742-6596/3025/1/012022
Ally, S., Verstraeten, T., Nowé, A., & Helsen, J. (2025). Optimal day-ahead trading and power control for a hybrid wind-hydrogen plant with multi-agent reinforcement learning. Journal of Physics: Conference Series, 3025(1), Article 012022. https://doi.org/10.1088/1742-6596/3025/1/012022
@article{50564d43e8b54588b13d0ebfd7b808d2,
title = "Optimal day-ahead trading and power control for a hybrid wind-hydrogen plant with multi-agent reinforcement learning",
abstract = "Hybrid wind-hydrogen plants have multiple revenue sources, subject to uncertainties and trade-offs. As a consequence, maximizing their overall profitability is a challenging optimization problem. Since electricity is typically traded in advance of its actual power generation, weather forecasts play a crucial role in the power trading strategy. Additionally, the trading and control strategies of other market players influence the imbalance of the public grid and, as a result, have an impact on the revenue that can be generated by grid balancing. To address this challenge, we propose a novel approach based on multi-agent reinforcement learning (RL), with two RL agents: one for day-ahead power trading, and a second for real-time power control of the wind farm and the electrolyzer. The RL system is trained using real-world data from a large offshore wind farm in the Belgian North Sea. Results demonstrate the effectiveness of this approach, with the RL agents collaboratively maximizing the total operational profit of the hybrid plant, achieving a significantly higher profitability compared to conventional methods.",
keywords = "hybrid wind-hydrogen plant, multi-agent reinforcement learning, power trading, power control, offshore wind farm",
author = "Stijn Ally and Timothy Verstraeten and Ann Now{\'e} and Jan Helsen",
note = "Publisher Copyright: {\textcopyright} Published under licence by IOP Publishing Ltd.",
year = "2025",
month = jun,
doi = "10.1088/1742-6596/3025/1/012022",
language = "English",
volume = "3025",
journal = "Journal of Physics: Conference Series",
issn = "1742-6596",
publisher = "IOP Publishing Ltd.",
number = "1",
}