State-of-health estimation for lithium-ion batteries based on electrochemical impedance spectroscopy measurements combined with unscented Kalman filter

Impedance-based state-of-health (SOH) estimation methods often rely on machine-learning algorithms to capture the complex correlation between the impedance and the capacity, requiring extensive offline characterization data. To address this, this study proposes a novel SOH estimation approach utilizing the unscented Kalman filter (UKF) to reduce the data requirement and improve robustness. Empirical models are trained offline to capture (1) the relationship between the SOH and the number of cycles, and (2) the relationship between the SOH and the health indicator, which is extracted via electrochemical impedance spectroscopy (EIS). The UKF determines online the relative weighting between the two models based on the real-time impedance measurement and produces a final SOH estimate accounting for both the degradation trend and the impedance. Three different model structures are considered and compared. Open-source datasets containing 51 cells following various aging-paths are used for validation. The estimator can reach an error between 1.4 to 1.7% SOH regardless of the state-of-charge (SOC) level. It shows robustness against the mismatch between the training and testing SOCs and can successfully extrapolated to previously unseen cell types. Its data efficiency is highlighted by its ability to maintain a low estimation error with insufficient training data.