State Of Charge Estimation For Lithium-Ion Battery Based On Improved Cubature Kalman Filter Algorithm

As an important part of new energy vehicle,the Battery Management System(BMS)has received extensive attention from researchers at home and abroad.Among the many research directions of BMS,how to accurately and efficiently estimate the state of charge(SOC)of lithium batteries is a hot topic of current research.In the process of SOC estimation using the Cubature Kalman Filter(CKF)algorithm,it is found that compared with the Extend Kalman Filter(EKF)and Unscented Kalman Filter(UKF),although CKF has a high estimation accuracy,there are still many constraints,for example,the covariance matrix is easy to lose positive definiteness,the filtering convergence speed is slow and the filtering accuracy is greatly affected by the measurement noise matrix.Focusing on the above issues,the specific work done in this paper consists of the following four aspects:Firstly,this article used capacity calibration experiment,constant current discharge experiment,HPPC(Hybrid Pulse Power Characteristic)experiment,DST(Dynamic Stress Test)experiment,UDDS(Urban Dynamometer Driving Schedule)experiment to test the battery at different temperatures,different discharge rates,and different dynamic conditions,then extracted charge and discharge data.The capacity,internal resistance and open circuit voltage characteristics of the experimental battery were also analyzed.Secondly,the Thevenin equivalent circuit model was selected to build the battery model.On the basis of this model,the values of ohmic resistance,polarization internal resistance,and polarization capacitance were identified.The accuracy of the built model was verified under constant current and DST conditions.Then,the covariance matrix diagonalization decomposition,strong tracking filter and improved noise adaptive algorithm were introduced to optimize the traditional CKF algorithm and the combined improved CKF algorithm was proposed.After applying EKF,UKF,CKF and improved CKF algorithms to constant current,DST and UDDS conditions,it was found that the improved CKF algorithm had higher estimation accuracy and stability whether in constant current condition or dynamic conditions.It also had an excellent performance in convergence and noise adaptation.Finally,in order to expanded the application range of the improved CKF algorithm to adapt to different working environments of lithium batteries,a variable forgetting factor recursive least square algorithm was introduced to identify the lithium battery parameter matrix online.The battery parameters were brought into the improved CKF algorithm to estimate the SOC after identified and updated in three special working conditions: low temperature,high temperature and aging conditions.The estimation results showed higher accuracy than the estimation results under the condition that the parameters were not updated.