Federated Learning Based State of Health Estimation for Lithium-Ion Batteries

Research on lithium-ion batteries focuses on parameters such as Remaining Useful Life (RUL), State of Health (SoH), and State of Charge (SoC). As batteries age, their capacity degrades, and their charging and discharging performance declines. SoH is a critical parameter that directly reflects the aging state of the battery and plays a significant role in maintenance processes. Therefore, real-time SoH monitoring in battery management systems (BMS) has become an important research area. In this study, SoH estimation using federated learning is proposed. The most critical moment for obtaining SoH information is at the end of the charging process. This information enables users to better plan which tasks the battery will be suitable for. Federated learning is a decentralized machine learning approach that processes data on local devices while aggregating model updates on a central server. This approach provides significant advantages in terms of privacy and data security. As a result of the study, the control parameters RMSE, MAE, and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{R}^{\mathbf{2}}$</tex> were calculated, yielding approximate values of 0.109, 0.086, and 0.994, respectively.