Lithium-ion Battery SOC Estimation Study Based On Improved EKF Algorithm

Kalman filter algorithm is a kind of recursive linear minimum variance estimation algorithm,which could get better estimation of the system state under the condition that the dynamic battery model was precise and the prior statistical characteristics of system noise were known.Since the battery is a complex nonlinear system,it is difficult to obtain an accurate battery model and the statistical properties of system noise in practical applications.Inaccurate battery model and inappropriate identification of the statistical properties of the system noise can result in reduced or even diverged precision of the filter.In order to improve the accuracy of SOC estimation and simplify the operation of the system,a SOC estimation method was proposed based on improved Extended Kalman Filter algorithm for lithium battery.Taking a 35Ah ternary-material battery as the research object,a detailed research was conducted on the battery model and the SOC estimation algorithm based on the proposed model.A second-order Thevenin equivalent circuit model was set up,then the parameters R0,R1,R2,C1,C2 of the battery model were identified by the pulse discharge experiment method and the recursive least squares identification method with genetic factors respectively.First,according to the principle of least squares algorithm,the parameters of battery models with different SOC were obtained by fitting the experimental data of pulse discharge method with Matlab programming.The mathematical models to characterize the dynamic characteristics of batteries were established with these parameters.Second,according to the recursive principle of the recursive least squares algorithm with genetic factors,the real-time battery model parameters were obtained through Matlab/Simulink modeling.Finally,based on the discharge HPPC experimental data,the model parameters identified by the two algorithms were verified in Matlab/Simulink environments,and the accuracy of the battery model and its adaptability to dynamic conditions was verified with the experimental data of the dynamic cycle of charging HPPC and DST.In addition,with the Ampere hour integral method,the SOC estimation of nonlinear battery system was realized with the help of the extended Kalman filter’s"prediction-correction-prediction" method.In order to further improve the precision of the battery model parameters and the state estimation,the model parameters identified by the pulse discharge method and the Dual Extended Kalman Filter Algorithm were both used to achieve a more accurate estimation of the battery model parameters and state.The traditional Extended Kalman Filter and Dual Extended Kalman Filter algorithm failed to accurately estimate the covariance of the system noise and observation noise,when estimating the battery SOC.This failure would lead to the decrease of filtering accuracy.The above-mentioned algorithm was improved with the correction principle of Cauchy robust function data.Based on the residual error between the predicted value and the actual value of the dual extended Kalman filter observation equation,the algorithm was made more robust by using the influence function to get a real-time correction to the state noise covariance matrix Q,and dynamically adjusting the observation noise covariance matrix R by the SOC interval.The algorithm not only improved the accuracy of the battery model,but also overcame the problem of the decrease of estimation accuracy due to the unknown statistical characteristics of the system noise in the extended Kalman filtering algorithm.The simulation results show that the estimation error of the improved extended Kalman filter algorithm was less than 2.5%,which verified the validity and accuracy of this method.