Green shipping via offshore wind-hydrogen integration: collaborative scheduling using multi-agent reinforcement learning

The integration of offshore wind power with green hydrogen production presents a promising pathway to enhance grid stability and provide zero-carbon fuel for maritime transport, thereby forming an integrated wind-hydrogen-shipping chain. However, large-scale implementation often faces challenges such as wind curtailment and uncertain demand from shipping. To overcome these issues and meet green shipping demands, this study proposes a collaborative scheduling framework for offshore wind-hydrogen coupling systems and hydrogen-fueled ships. Firstly, an event-driven mechanism is introduced to construct a dynamic priority scheduling model for hydrogen-fueled ships. The framework integrating continuous energy scheduling with discrete event-driven mechanisms has been established. Secondly, a collaborative decision-making architecture based on multi-agent deep deterministic policy gradient algorithms is developed through the generation of dual uncertainty scenarios. The collaborative optimization is achieved through centralized training and decentralized execution. Results shows that the proposed framework and algorithm enhance daily profit and hydrogenation efficiency while reducing curtailment rates. This validates the framework’s effectiveness and innovativeness in addressing source-load uncertainties and optimizing the spatiotemporal allocation of hydrogen energy. It provides decision-making paradigms for the collaborative development of offshore integrated energy hubs and green shipping.