State-of-charge estimation of Lithium-ion batteries using an adaptive dual unscented Kalman filter based on a reduced-order model

It is critical to accurately estimate the state of charge (SOC) of Lithium-ion batteries for the efficient, reliable, and safe operation of Electric Vehicles (EVs). Therefore, it is crucial to improve the methods for SOC determination by finding a beneficial trade-off between higher accuracy and increasing complexity within the Battery Management System (BMS). Physics-based models outperform the Equivalent Circuit Model (ECM) in terms of inherent electrochemical representation. This paper develops a new model based on the derivation of the reduced-order battery model in a control-oriented state space model to incorporate with the adaptive unscented Kalman filter (ADUKF). This allows the BMS to use a physics-based model with a low number of battery parameters resulting in a complexity deduction and much less effort in electrochemical parameter identification for online co-estimation of SOC and the battery model parameters. Moreover, an adaptive dual spherical UKF is implemented with the forgetting factor to reduce the computational capacity required for the traditional UKF and also to improve the covariance matching of measurement and system process noises. The proposed method is validated against driving cycle profiles, considering various conditions such as erroneous initial SOC, inaccurate battery capacity, and current sensor inaccuracy. Further, the robustness of the proposed method is verified against the impact of different temperatures. Lastly, the result is compared to ADUKF based on ECMs in terms of accuracy and computational cost.