Online parameters identification and state of charge estimation for lithium-ion batteries based on improved central difference particle filter

Abstract

Accurate SOC is of great significance to alleviate the endurance anxiety of electric vehicle drivers. To improve the accuracy of SOC estimation, a method of combining FFRLS and GPR for parameter identification and an improved particle filter for lithium battery SOC estimation are proposed. The proposed collaborative parameter identification method uses the forgetting factor recursive least squares (FFRLS) to identify the parameters in the middle and high SOC region. In the low SOC region, the parameters identified by the FFRLS are used as the training data of Gaussian process regression (GPR) for model parameter prediction to improve the parameter identification accuracy in the low SOC region; in the SOC estimation stage, the central difference Kalman filter is used to generate the importance density function to overcome the particle degradation, and the step size h of the central difference is calculated suboptimal to further improve the prediction accuracy. Finally, the proposed method is compared with other methods under different working conditions and different temperatures, and the effectiveness and adaptability of the proposed method are verified.