Research On The Joint Estimation Of Lithium-ion Battery State Of Charge And Capacity

As the core energy source of electric vehicles,power batteries directly restrict the development of electric vehicles,and accurate estimation of SOC is not only the fundamental function of electric vehicle battery management system,but also helps to improve energy utilization of battery,safeguard the application of batteries in EVs,extend the cycling life.However,the time-varying nonlinearity,environmental sensitivity,and irreversible decay during use of the battery make the estimation of hidden states such as SOC a challenge to the industry.This study conducted the following research on the SOC and capacity estimation of lithium-ion batteries:(1)Aiming at the parameter identification problem of battery models,3 discrete mathematical models(n= 0,1,2)based on the n-RC equivalent circuit model are established,the off-line parameter identification process choose Genetic Algorithm,at the same time,the parameter on-line identification use the recursive least squares method with forgetting factor.Evaluate the performance of the three models from the aspects of model identification accuracy and model calculation,the results show that the comprehensive performance of the 1-RC model is optimal.(2)To achieve the co-estimation of battery’s state and parameters,an adaptive cubature Kalman filter SOC estimation method based on random weighting(ARWCKF)is proposed,at the same time,use extended Kalman filter to identify the parameter on-line.which,significantly improve the robustness and accuracy of the algorithm.Through the prediction accuracy of terminal voltage,estimation precision of SOC,and the convergence rate with inaccurate initial value to compare the algorithms with FRLS-EKF,FRLS-CKF and EKF-CKF.The results verify that this approach has a better performance with the error of SOC is under 3%.(3)Aiming at the limitations of the single time-scale joint estimation algorithm,the time scale of parameter identification is adaptively adjusted to achieve coestimation of battery capacity and SOC,based on the EKF-ARWCKF joint estimation algorithm with variable time scale,which uses macro time scale to identify battery capacity and model parameters,the micro time scale to estimate SOC,and take accumulated discharge as the conversion standard between micro and macro time scales.The filtering performance of the algorithm is effectively evaluated based on the prediction accuracy of the terminal voltage,SOC,capacity and the convergence rate of SOC and capacity,verifying that compared to the single-timescale approach this approach has a better estimation performance.(4)For the purpose of the practical application,a hardware-in-the-loop simulation test verification platform is established based on the A&D5435 platform,power battery test system.Using MATLAB/Simulink software to build the battery capacity and SOC joint estimation model,and then load it into A&D54354 platform for further research.The results indicate the feasibility and robustness of the proposed approach even at inaccurate initial SOC and capacity,which has certain practical application value.