Xiaofen Fang, Weidong Chen, Tao Hu, Zijie Chen, Qingqiang Zeng, Jianqing Li
{"title":"Federated learning empowered microgrids: Lithium battery state-of-health prediction and multi-node co-optimisation","authors":"Xiaofen Fang, Weidong Chen, Tao Hu, Zijie Chen, Qingqiang Zeng, Jianqing Li","doi":"10.1016/j.est.2025.118892","DOIUrl":null,"url":null,"abstract":"<div><div>With the wide application of lithium-ion batteries in microgrids, accurately predicting the State of Health (SOH) of the batteries and realising multi-node co-optimisation has become a key challenge to ensure system stability and prolong battery lifetime. In this paper, we propose a novel federated learning-driven SOH prediction and multi-node optimisation model for Li-ion batteries, named FedOptSOH. The framework is based on National Aeronautics and Space Administration (NASA)’s publicly available Li-ion battery dataset, and utilises heterogeneous data distributed across nodes to achieve highly accurate SOH prediction through local training and central server aggregation, while effectively protecting data privacy. FedOptSOH integrates advanced federated optimisation algorithms, Federated Average of Momentum (FedAvgM) and Adaptive Federated Optimisation (FedOpt), which significantly improves the training stability and convergence speed of the model. Based on the predicted SOH results, FedOptSOH further employs a multi-objective co-optimisation approach to achieve dynamic adjustment of charging and discharging strategies at each node in the microgrid, maximising energy efficiency and delaying battery degradation. Experimental results show that FedOptSOH achieves an Mean Absolute Error (MAE) of 4.17%, a Root Mean Square Error (RMSE) of 5.03%, and a coefficient of determination (R<span><math><msup><mrow></mrow><mrow><mn>2</mn></mrow></msup></math></span>) of 0.978 in SOH prediction, which are significantly better than the traditional methods; multi-node co-optimisation effectively reduces the energy loss of the system by 12%, and extends the battery life by 15%. FedOptSOH framework combines high accuracy, strong privacy protection and real-time co-optimisation, providing a solid theoretical foundation and technical support for lithium battery management in smart microgrids.</div></div>","PeriodicalId":15942,"journal":{"name":"Journal of energy storage","volume":"139 ","pages":"Article 118892"},"PeriodicalIF":8.9000,"publicationDate":"2025-10-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of energy storage","FirstCategoryId":"5","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2352152X25036059","RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"ENERGY & FUELS","Score":null,"Total":0}
引用次数: 0
Abstract
With the wide application of lithium-ion batteries in microgrids, accurately predicting the State of Health (SOH) of the batteries and realising multi-node co-optimisation has become a key challenge to ensure system stability and prolong battery lifetime. In this paper, we propose a novel federated learning-driven SOH prediction and multi-node optimisation model for Li-ion batteries, named FedOptSOH. The framework is based on National Aeronautics and Space Administration (NASA)’s publicly available Li-ion battery dataset, and utilises heterogeneous data distributed across nodes to achieve highly accurate SOH prediction through local training and central server aggregation, while effectively protecting data privacy. FedOptSOH integrates advanced federated optimisation algorithms, Federated Average of Momentum (FedAvgM) and Adaptive Federated Optimisation (FedOpt), which significantly improves the training stability and convergence speed of the model. Based on the predicted SOH results, FedOptSOH further employs a multi-objective co-optimisation approach to achieve dynamic adjustment of charging and discharging strategies at each node in the microgrid, maximising energy efficiency and delaying battery degradation. Experimental results show that FedOptSOH achieves an Mean Absolute Error (MAE) of 4.17%, a Root Mean Square Error (RMSE) of 5.03%, and a coefficient of determination (R) of 0.978 in SOH prediction, which are significantly better than the traditional methods; multi-node co-optimisation effectively reduces the energy loss of the system by 12%, and extends the battery life by 15%. FedOptSOH framework combines high accuracy, strong privacy protection and real-time co-optimisation, providing a solid theoretical foundation and technical support for lithium battery management in smart microgrids.
期刊介绍:
Journal of energy storage focusses on all aspects of energy storage, in particular systems integration, electric grid integration, modelling and analysis, novel energy storage technologies, sizing and management strategies, business models for operation of storage systems and energy storage developments worldwide.