Research On State Of Health Estimation Method Of Lithium-ion Battery

Li-ion battery is an important part of energy storage system,and need to be properly managed to improve reliability,safety and prolong service life.Since Li-ion batteries are nonlinear devices with complex electrochemical structures,and their performance is largely influenced by internal and external factors(aging,temperature),steady and accurate prediction of lithium battery health is helpful in predicting working range,protection from potential damage and improved performance.This text takes the NCR18650 PF lithium ion battery as the test object,and focuses on the actual-time prediction algorithm of the SOH for lithium battery on the basis of building a battery model and parameter identification.Firstly,the existing battery models are compared and analyzed from the aspects of model complexity and estimation accuracy.Based on the comprehensive consideration of the analysis results,the equivalent circuit model is determined as the basis of subsequent estimation,so as to establish the system state of the battery model equations and observation equations.Based on the power battery test system connected with various devices,various performance tests and DST dynamic working condition tests are carried out on the tested battery.Taking into consideration that the intrinsic parameter of the model are slowly time-varying,an IDOA optimization algorithm based on macro time scale is proposed to identify it offline,and it is verified by DST dynamic conditions.Secondly,considering the nonlinear characteristics of lithium batteries,the SOC prediction of lithium batteries is realized based on the unscented Kalman filter algorithm(UKF).In view of the fact that the algorithm will stop running when it encounters an error covariance matrix that is not positive definite,singular value decomposition(SVD)is adopted to deal with it.Since the algorithm is running with fixed state noise and observation noise,the prediction of SOC will produce continuous cumulative errors.Based on this,an adaptive filtering algorithm is integrated to form an improved UKF algorithm.However,in the actual process of lithium batteries,the initial value of SOC is unknown.In order to improve the estimation robustness of the UKF algorithm,the maximum entropy criterion was introduced,and the MC-IUKF algorithm was proposed to estimate the SOC,and compared through the DST dynamic working condition test.Verification,the results show that compared with the UKF algorithm,the MCIUKF algorithm has a faster correction ability and follow-up ability to the initial value error.Then,based on the charging and discharging characteristics of lithium battery,ohmic resistance is finally established as the characterization parameter of battery SOH.Considering the dynamic characteristics of model parameters,EKF algorithm is introduced on the basis of MC-IUKF,which realizes the real-time update and prediction of ohmic internal resistance and SOC.In the calculation formula of the internal resistance definition method,the SOH estimation of the lithium battery is realized.Finally,the test plan of UDDS working condition(urban road cycle working condition),US06 working condition(severe driving working condition)and FUDS working condition(federal city running working condition)was constructed and tested through the power battery test system.The SOH estimation results obtained by the EKF-UKF algorithm and the EKFMC-IUKF algorithm show that the improved algorithm not only has higher prediction robustness than the standard algorithm,but also is almost unaffected by the initial value,and has a fast approaching speed.Continue to use EKF for FUDS conditions-MC-IUKF algorithm is used to estimate SOH,and it is verified that the proposed EKF-MC-IUKF algorithm has strong adaptability and high estimation accuracy under different working conditions.