Study On Modeling And SOC Estimation Of Lithium-ion Power Battery For Vehicle

As an indispensable component of new energy vehicle technology,Battery Management System(BMS)has been widely concerned by domestic and foreign researchers.With in-depth research of BMS,the accuracy of battery state estimation is improved gradually,but the structure of battery model has also become more complicated.In order to solve the balance problem between the battery model structure’s complexity and the prediction accuracy,this paper judged the optimal model of the ternary lithium ion battery based on Akaike’s Information Criterion(AIC).Considering the time-varying characteristics of the model parameters and the battery aging characteristics,this paper improved the accuracy of State of Charge(SOC)of the parallel battery by studying the key technology of joint estimation of the battery model parameters and SOC.This article mainly researched on four aspects,namely the basic characteristics of the ternary lithium ion power battery,the study of the optimal model of the single battery,the study of the optimal model of the parallel battery and the joint estimation method of the model parameters and SOC based on the EKF-CKF.The specific work is as follows:First of all,an in-depth research on the basic characteristics of ternary lithium-ion batteries was conducted.Taking the power battery with a rated capacity of 40 Ah produced by a company in Jiangsu as the research object,conducting capacity calibrating and cyclic aging condition test,Constant Current Discharge(CCD)working condition test,Hybrid Pulse Power Characteristic(HPPC)working condition test,Dynamic Stress Test(DST)working condition test,the main discharge working condition test of the energy battery for pure electric passenger vehicles,Hall current sensor calibration test and the parallel battery test.Based on the experimental data,the capacity characteristics,voltage characteristics and internal resistance characteristics of the battery were analyzed.Then,based on the n-order RC equivalent circuit model,a comprehensive performance function of structure complexity of quantitative evaluation model and prediction accuracy were proposed to judge the optimal model of the lithium ion single battery by AIC.Through offline and online modeling methods,AIC analysis was performed on equivalent circuit models of different RC orders,and it was found that the one order RC equivalent circuit was the best performing model in describing the characteristics of ternary lithium-ion batteries.On this basis,this paper proposed that processing the voltage and branch current in AIC residual data using min-max normalization,and proposed a comprehensive performance function of the complexity and prediction accuracy of the parallel battery model in the sense of time-frequency domain tuning by modifying the Akaike’s information criterion to determine of the optimal model order of parallel battery.Then,the applicability of the optimal model was verified under the dynamic cycling condition and the extreme difference state of the battery.Finally,based on the study of the basic characteristics of the battery and the optimal model,a state space equation was established based on the one order RC equivalent circuit model.The results of SOC estimation of the Extended Kalman Filter(EKF),the Unscented Kalman Filter(UKF),Cubature Kalman Filter(CKF)algorithms respectively,and it was proved that the CKF algorithm has higher accuracy.Furthermore,an EKF-CKF joint estimation algorithm considering the time-varying characteristics of the model parameters and the aging characteristics of the battery itself was proposed for SOC estimation of the parallel battery.The accuracy and stability of the EKF-CKF algorithm were verified under the main discharge working condition...