Research On SOC Estimation Method Of Power Lithium Battery Cell Based On Kalman Filter

In recent years,electric vehicles have developed rapidly.BMS(Battery Management System)is very important during the operation of electric vehicles.As the basis of BMS,the battery's SOC(state of charge)can intuitively reflect the remaining capacity of electric vehicles,indirectly indicate the remaining cruising range of electric vehicles,and at the same time affect the overall vehicle control strategy of electric vehicles.However,SOC is not a directly measurable quantity.Battery charging and discharging involve complex chemical and physical processes,and the battery system has time-varying and nonlinear characteristics.In addition,the vehicle-mounted environment is complex and harsh,and it is often impossible to accurately collect data.Therefore,it is necessary to adopt an appropriate method to estimate the SOC.In this thesis,the research on the estimation method of SOC of vehicle power lithium battery starts with the characteristic test of power lithium battery,then builds a second-order RC equivalent circuit model of the battery based on the test results,and finally designs the SOC estimation method based on the model and performs verification analysis.This thesis mainly considers the influence of temperature and discharge rate in modeling,and overcomes the limitation of single temperature or single discharge rate in the previous research;in the improvement of algorithm,this thesis designs the joint estimation of Kalman Filter and Ampere-hour Integration.The SOC solution effectively solves the problem of insufficient sensor accuracy in actual situations.The main research contents of this thesis include:1.Battery test.Most of the existing studies only consider a single influencing factor,and other influencing factors are often treated by the method of correction coefficient.In this thesis,the NCR18650 PF battery is selected as the experimental object,a battery test program that considers both temperature and discharge rate is designed,and the battery capacity test,pulse discharge test and OCV-SOC curve test at different temperatures and different discharge rates are completed.And this paper analyzes the battery experimental data,which provides a basis for battery modeling and model parameter identification.2.Battery modeling.In this thesis,combined with the experimental data of battery testing,on the one hand,the battery capacity at different temperatures and different discharge rates is calibrated.On the other hand,the battery model parameters at different temperatures and different discharge rates are identified,and the battery modeling is completed.It has been verified by simulation that under constant current discharge,pulse discharge conditions and The Hybrid Pulse Power Characterization(HPPC)conditions,the voltage error of the output terminal of the battery model in the pre-discharge and mid-discharge period is within 0.05 V.Under New European Driving Cycle(NEDC)conditions,due to frequent current switching,the model error reaches the 0.4V,the average error is still within 0.05 V,indicating that the battery model established in this paper has high accuracy.3.Improvement of SOC estimation algorithm.In this thesis,for the battery model in the late stage of the discharge,the chemical reaction is unstable and the error increases.Based on the Extened Kalman Filter(EKF)algorithm,the self-adaptation of noise is considered,and the SOC estimation is combined with the Ampere-hour Integration at the later stage of discharge.And the simulation experiment of the SOC estimation scheme under constant current and pulse discharge conditions is completed.And the average error of the estimation is within 1%.In the simulation of two charging conditions including HPPC and NEDC,the average error of SOC estimation is within 4% and 8% respectively,indicating that the SOC estimation scheme in this thesis can estimate the accurate value of SOC well.