State Of Charge Estimation Of Lithium-ion Battery Based On UKF Method

With the popularity of the electric vehicles in recent years,the development and optimization of battery management system has become a topical issues in the automotive industry.Among the parameters measured by BMS,the state of charge(SOC)of battery is an important index which can measure the residual capacity of battery.Since it can not be measured directly by instrument,only can use the external characteristic parameters such as working voltage and current of power battery and appropriate estimation algorithm to estimate it.Therefore,accurately estimating the SOC of power battery is the focus of research in the application of electric vehicle.Firstly,this paper has introduced the development prospects of electric vehicles and power batteries,and compared the advantages and disadvantages of several common lithium-ion batteries in the market.the single lithium-ion phosphate power battery is selected as the experimental object.Because of the internal parameters of lithium-ion batteries can not be measured directly,only the voltages at both ends of the batteries during charging and discharging current can be measured,and the changing rules of the internal parameters of the batteries can be obtained by changing the voltages at both ends of the batteries.Therefore,a large number of charging and discharging experiments are carried out to find the internal variation of lithium-ion batteries from the constantly changing terminal voltage.Secondly,accurate parameters are the basis of accurately predicting the SOC of lithium-ion batteries.By comparing several commonly used battery models,the second-order RC loop model which is most in line with the characteristics of the battery is taken as the battery model in this paper.the equation of the battery model is established according to Thevenin’s theorem.Through current pulse charging and discharging experiments on lithium-ion batteries,the rebound curve of terminal voltage is obtained,and the self-defined equation is fitted,and the variation rules of each parameter in the battery model are obtained.Experiments show that the internal parameters of batteries are not fixed during the charging and discharging process.They change with the change of battery capacity.In this paper,these changes are sorted out,and these real-time parameters are substituted into the battery model to improve the accuracy of battery SOC prediction.Finally,the state equation and output equation of the battery model are established,and the estimation of SOC of the battery is studied by combining the unscented Kalman filter method with the traditional current integration method and open-circuit voltage method.Based on the predicted results of battery SOC by current integration method,the results of unscented Kalman filter method are compared with those of extended Kalman filter method which is often studied in recent years.Because the unscented Kalman filter uses unscented transform(UT)to process data,it does not need to linearize the non-linear equation,and avoids the error caused by linearization.The error of parameter estimation is smaller than that of extended Kalman filter,and the convergence effect is more obvious than that of extended Kalman filter.The unscented Kalman filter method is applied to simulated working conditions of variable current,and the estimation results are compared with those obtained by the current integration method,which also achieves good results.