Joint estimation of state-of-charge and state-of-health for all cells in the battery pack using “leader-follower” strategy

Developing simple and accurate state estimators for battery packs is important but technically challenging due to not only the high number of batteries requiring monitoring but also the uncertainties brought by the inherent cell inconsistency and the additional equalisation hardware. We here propose a low-computational “leader-follower” framework to achieve state-of-charge (SoC) and state-of-health (SoH) estimations for all series-connected cells within a pack. It basically uses an enhanced algorithm to handle a selected battery (“leader”) and updates the states of the remainders (“followers”) with lightweight calibrators. Specifically, a revised extreme-learning machine equipped with two gradient correctors is first developed to estimate the SoC of the “leader” adaptively, followed by a simple yet effective definition-based approach for its SoH. The states of the “followers”, on the other hand, are calibrated based upon the relationships among voltage, SoC, and a concept called balancing-current-ratio (BCR). Battery-in-the-loop experiments show that when the computational time is reduced by 83% compared to the benchmarks, the typical estimation error of all cells in a pack can still be bounded by 2.5% and 1.25% for SoC and SoH, respectively. Given the low computational burden of our algorithm, it can be easily transplanted to applications with different system configurations.