Research On SOC Estimation Of Lithium-ion Battery For Electric Vehicles

For the safe and efficient operation of electric vehicles,the battery management system must be able to accurately estimate the State of Charge(SOC).In this paper,Nickel-Cobalt-Meng ternary lithium-ion battery is taken as the research object,and electric vehicle is taken as the application background.Aiming at the accuracy of battery SOC estimation,a reasonable battery model is established.The SOC estimation algorithm is discussed with emphasis to realize the online estimation of battery SOC.Firstly,a battery simulation model was established.For the problem that battery parameters change with the operating environment and current conditions,the Thevenin model,which is updated dynamically with the change of temperature and SOC,is established.The parameters in the model are expressed by a two-dimensional lookup table.To obtain the two-dimensional look-up table of parameters,constant current discharge experiments were carried out on batteries at different ambient temperatures,and the parameters of the model were estimated by offline parameter identification method in MATLAB using the experimental data.Based on the simulation model of batteries,SOC estimation based on adaptive unscented Kalman filter(AUKF)is studied.In the estimation of SOC,the state noise and measurement noise are usually given as a preset value.However,the power battery of electric vehicle is a time-varying dynamic system in practical application,and the uncertainty of noise value will lead to the increase of estimation error.Therefore,combining the adaptive covariance algorithm with the unscented Kalman algorithm,an improved adaptive unscented Kalman filter algorithm is studied to correct the state noise and measurement noise in real time,which reduces the estimation error of SOC under dynamic system noise.In order to verify the estimation effect of AUKF,SOC estimates are tested under constant current discharge in lab,US06,UDDS and HW-FET conditions in ADVISOR environment.The accuracy,convergence and robustness of AUKF are verified.The results show that the proposed SOCestimation method can realize the SOC estimation of batteries and has high estimation accuracy in actual driving conditions of electric vehicles.Finally,aiming at the problem of capacity degradation in the actual use of batteries,the recursive approximate weighted TLS(AWTLS)algorithm is used to estimate the capacity of batteries.The accuracy of the AWTLS algorithm to estimate the maximum available capacity is verified under constant current discharge conditions and random conditions.The simulation results show that the AWTLS algorithm can accurately estimate the battery capacity,and the estimation results can improve the estimation of SOC.