State of Charge Estimation of Lithium-Ion Batteries with Adaptive Square Root Central Difference Kalman Filter

Abstract

Lithium-ion batteries have been a major energy source in electric vehicles because of their strong adaptability in operating conditions. Accurate estimation of state of charge (SOC) for lithium-ion batteries can efficiently improve the efficiency of battery energy utilization. However, SOC estimation is complicated in operating conditions with unknown model parameters. This study proposes an adaptive square root central difference Kalman filter (ASRCDKF) algorithm based on the equivalent circuit model to achieve high-precision estimation of SOC. First of all, to avoid an open-circuit voltage test, a linear Kalman filter is constructed to realize real-time estimation of unknown parameters in the measurement equation. Then, to improve the stability of the algorithm, a square root method is used to ensure a positive semi-definite of the error covariance matrix that is based on the adaptive central difference Kalman filter algorithm. Model parameters are considered as the state to be estimated, and the joint estimation of the model parameters and SOC is realized by an ASRCDKF algorithm. After that, the linear Kalman filter is coupled with the ASRCDKF to realize the accurate estimation of SOC in the case of both the state equation and the measurement equation including unknown parameters. Last, the ASRCDKF algorithm is compared with the adaptive central difference Kalman filter algorithm and the adaptive cubature Kalman filter algorithm under two sets of operating conditions. The results show that the SOC estimation of the ASRCDKF algorithm is more significantly accurate than other algorithms under different operating conditions.