Research On Robustness Estimation Of The State Of Charge For Lithium-ion Batteries

Robustness estimation of State of Charge(SOC)of the li-ion battery was studied for electric vehicles based on the battery model in this paper.Estimation of SOC are meaningful to safety management of the battery,estimation of the remaining mileage etc.Therefore,following works was done:First of all,two kinds of lithium batteries were tested in different temperature conditions and different discharge conditions,whose charge and discharge characteristics were compared.Then,Least squares method(LS),forgetting factor least squares method(FRLS),the bias compensation least squares method(BCRLS)were used to realize the off-line and on-line parameter identification.The results show that the on-line identification is more suitable for the dynamic process of the battery than the off-line identification.FRLS and BCRLS have similar identification accuracy,but BCRLS is better for colorless noise.Then,the extended Kalman filter(EKF),the H_? filtering method(HIFF)and the joint estimation of parameters and states based on the online parameter identification were studied.The accuracy of the above algorithms is verified by DST and UDDS test.The results show that the EKF method has better accuracy for white noise.Although the HIFF method is conservative,the robustness is good for all kinds of colored noise and rough model.Finally,the redundant design scheme of SOC estimation was studied.In order to overcome the contradiction between the single model,the accuracy and stability of the algorithm,an SOC fusion estimation algorithm with multi-model and multi-algorithm was designed.The results show that the fusion algorithm can improve the accuracy and stability of the algorithm,and strengthen the applicability under different working conditions and redundancy of the algorithm.In general,the fusion algorithm has a higher accuracy and stronger robustness to adapt to complex conditions.