A Distributed Multi-microgrid Intelligent Scheduling for New Power System

With the new power system growth, cooperative scheduling among multiple microgrids (MMG) is emerging. Complex energy coupling relationship within the MMG system poses challenges for achieving energy complementarity. Tradition-al MMG scheduling faces limitations in achieving economic scheduling and privacy protection due to the extensive need for energy data. To address the MMG scheduling issue, this paper proposes a novel distributed intelligent cooperative scheduling model, named E-Hive, for optimal economic operation. We employ two techniques into E-Hive: 1) We develop a multi-agent deep reinforcement learning module to achieve cooperative scheduling in MMG. 2) We employ a distributed architecture for inter-microgrid communication while ensuring privacy protection of energy data. Evaluation results show that the E-Hive model enables cooperative scheduling relying solely on local microgrid data, preserving the privacy of each microgrid. Furthermore, the operating cost is reduced by up to 16.4% compared to state-of-the-art methods, enhancing the economic benefit.