Model-based state of charge estimation for LFP cells: Enhanced ECM & EKF performance and robustness analysis
Files
PASTURE_57252100_2026.pdf
Closed access - Adobe PDF
- 4.78 MB
Details
- Supervisors
- Faculty
- Degree label
- Abstract
- Accurate State of Charge (SoC) estimation is a critical function of Battery Management Systems (BMS) in industrial Battery Energy Storage Systems (BESS). As SoC cannot be measured directly, it must be inferred from sensor data using models and estimation algorithms. This work develops and evaluates a model-based SoC estimation approach based on an Extended Kalman Filter (EKF), in collaboration with ENGIE Laborelec. The proposed method combines an equivalent circuit model (ECM) with real-time measurements in a prediction–correction framework. Model improvements, including second-order dynamics and extended parameter dependencies, are identified and validated on laboratory datasets before integration into the estimator. The study is conducted at the cell level, which serves as the fundamental building block for higher-level BMS control. The resulting approach significantly improves estimation accuracy, reducing RMSE by up to 58.6% on similar SoH datasets and 28.2% across all datasets, reaching final errors of 1.31% and 1.91%, respectively. The analysis also evaluates robustness through statistical consistency and parameter sensitivity. Results show that a well-tuned fixed EKF provides more reliable and robust performance than an adaptive filtering. In addition, they reveal a fundamental limitation of EKF for SoC estimation: they do not provide a reliable estimate of uncertainty due to the violation of its Gaussian assumptions. Sensitivity analysis further identifies the most critical parameters for estimation accuracy, providing practical guidelines for BMS design. These findings provide a clear and practical framework for SoC estimation in industrial BESS, highlighting both effective solutions and key limitations, and offering actionable insights for deployment at Laborelec.