Field-Validated Battery Capacity Estimation Using an Iterative Filter-Based Approach

Accurate battery capacity estimation remains challenging due to the diverse operating conditions. This study proposes an iterative, filter-based estimation framework that improves the accuracy and robustness of conventional approaches. The proposed estimator adaptively combines two features - the total energy consumption since the beginning-of-life and a health indicator (HI) extracted from the charging curve - with an unscented Kalman filter (UKF). The estimator is applied to a two-year field dataset collected from 20 electric vehicles to estimate the battery capacity on a monthly basis. The proposed iterative estimator can achieve lower estimation error in all cases compared to conventional approaches, especially when noises are introduced into the HI measurements. The results highlight the improved robustness and stability of the iterative estimator, with future work aiming to incorporate nonlinear models and larger datasets.