Concurrent estimation of lithium-ion battery charge and energy states by fractional-order model and multi-innovation adaptive cubature Kalman filter

To address the difficulties of the joint SOC and SOE estimation methods in achieving high accuracy and low complexity, a fractional-order multi-innovation adaptive square root cubature Kalman filter (FMASR-CKF) is proposed for the first time in this study. Firstly, a second-order fractional-order model (FOM) is established, and an improved dynamic genetic particle swarm optimization (DGPSO) algorithm is proposed for parameter identification. Then, the theory of multiple innovations is introduced into CKF to realize multi-step prediction based on historical data, thus enhancing the accuracy and robustness of the algorithm. At the same time, a joint estimation framework is established to correct and accurately estimate the SOE in real time through a simple relational equation. Validation under a variety of complex operating conditions shows that the mean absolute error (MAE) of the FMASR-CKF estimates of SOC and SOE is less than 0.5 % and 0.7 %, respectively. At 25 degrees C Federal Urban Driving Schedule (FUDS), the root mean square error (RMSE) for the SOC and SOE are 0.35 % and 0.62 %, respectively. Therefore, the proposed method exhibits high accuracy and robustness under a variety of real-world conditions with low complexity, providing an effective reference for the practical application of BMS.