Robust Parameter Identification For Equivalent Circuit Model And SOC Estimation Of Lithium Battery

With the increasing consumption of traditional fossil energy and the increasing efforts of governments in environmental protection,electric vehicles(EVs)powered by new energy have been valued worldwide because of their low energy consumption and pollution-free advantages.As one of the most important core technologies in the battery management system of electric vehicles,lithium battery State of Charge estimation plays an important role in the safe operation and energy-saving management of the vehicle.When the electric vehicle runs in complex conditions,how to accurately identify the battery model parameters online and how to improve the existing estimation algorithm to deal with non-Gaussian noise is still one of the research difficulties.Therefore,the design of robust identification algorithms and State of Charge estimation algorithms still have important academic value.This paper firstly analyzes the second-order equivalent circuit model(ECM)of lithium battery,determines the battery model measurement equation and observation equation and writes the battery SoC-OCV equation.Secondly,in order to effectively suppress the influence of noise in the model on the identification accuracy,this paper adds a bias compensation term in the recursive least squares algorithm to make it more accurate to identify the modelparameters in the presence of noise.Thirdly,considering the non-Gaussian noise as the main interference of the system,this paper uses the maximum correlation entropy criterion to improve the objective function of Extended Kalman Filter(EKF),and derives the Entropy-Extended Kalman Filter.(C-EKF)algorithm.In order to enhance the digital stability of the C-EKF algorithm,the C-EKF algorithm is improved by Weighted Least Squares,and finally the extended correntropy Kalman filter algorithm(C-WLS-EKF)based on the Weighted Least Squares is obtained.The simulation case analysis shows that the improved parameter identification algorithm and the State of Charge estimation algorithm can meet the requirements of online identification and achieve optimal estimation under the interference of non-Gaussian noise.