Accurate state-of-charge estimation of LiFePO4 battery: An adaptive extended kalman filter approach using particle swarm optimization

Ensuring the safety and reliability of LiFePO 4 battery relies heavily on accurate state of charge (SOC) estimation. The Extended Kalman Filter (EKF) algorithm is widely used for SOC estimation, but this method struggles to distinguish subtle open circuit voltage (OCV) variations and often misinterprets them as noise. This poses a challenge to the methodology for estimating SOC based on OCV. In this paper, an innovative method is presented to overcome this issue and enhance SOC estimation accuracy during the OCV plateau period. This method introduces an adaptive gain in the EKF, which is specifically designed for the OCV plateau period. To optimize the parameters of the adaptive gain function and improve the convergence performance of the estimator, Particle Swarm Optimization (PSO) is employed. By adapting the Kalman gain dynamically with this adaptive gain, the EKF effectively rebalances the confidence level between prior estimation and measurement, which can reduce the impact of OCV noise on SOC estimation and improve accuracy significantly. Extensive simulation experiments validate the practicality and effectiveness of this method and demonstrate its ability to enhance SOC estimation accuracy for LiFePO 4 battery during the OCV plateau period. Compared with the traditional EKF, the maximum error of this method does not exceed 2 %, and the average MAE is reduced by 27.64 % and the average MSE is reduced by 39.24 %. • An adaptive extended Kalman filter based on particle swarm optimization is proposed for LiFePO4 battery SOC estimation. • The impact of observation noise during open circuit voltage plateau period on SOC estimation accuracy is investigated. • Experimental results verified that the variation of Kalman gain can effectively enhance the SOC estimation accuracy.