Scientific Development Research

Aiming at the wide application of distributed energy storage systems in power grids, this paper proposes a SOC consistency control method based on federated learning. Through the federated learning framework, each distributed energy storage unit can achieve efficient state monitoring and collaborative control without sharing data. This method realizes precise control of SOC consistency by jointly training the battery states of multiple energy storage units, reducing system losses and performance degradation caused by unbalanced power levels of individual energy storage units. Comparative experiments in this paper verify the effectiveness and robustness of the proposed method in battery management systems. It has strong scalability and adaptability, and is suitable for the control requirements of distributed energy storage systems in future smart grids.

Distributed energy storage; Federated learning; SOC consistency; Battery management; Smart grid