Study Of Internal States Estimation Of Power Batteries Based On Equivalent Circuit Models For Electric Vehicles

The development of power batteries directly determines the prospects of electric vehicles(EVs),for the system of power potteries is the main energy source of EVs.The performance of power batteries not only lies on the materials and manufacture of the cell,but also is intensively related to the management system.The estimation of battery state of charge(SOC)has been a technical bottleneck for battery management system.Since the power battery application environment is complex and its remaining power can't be measured directly by the sensing device,the battery model is needed to achieve the goal of SOC estimation.In this study,a ternary material lithium-ion battery is selected to study the dynamic estimation of the battery model and SOC.The main contents are described as follows.Firstly,the structure and the principle of several common lithium-ion batteries are described.A ternary material lithium-ion battery was selected as the research object.The basic characteristics of the battery such as voltage,internal resistance and capacity under different discharge rates and temperatures were analyzed.The results show that under different discharge rates and temperatures,the changes of the terminal voltage and the capacity are obvious.The fuction of open circuit voltage and SOC always keeps stable in different temperatures,while the ohmic resistance significantly increased in low temperature environment.Secondly,the widely used extended Kalman filter(EKF)algorithm is selected to estimate SOC on the basis of the Thevenin,PNGV and DP battery model,respectively.The model accuracy,SOC accuracy and the quantitative relationship between the accuracy of the battery model and SOC are analyzed under three different operating conditions.The results show that the DP model has the highest precision and the strongest dynamic adaptability.The SOC estimation results based on the three equivalent circuit models can adapt to a variety of test conditions,of which the SOC estimation accuracy based on the DP model is the best.The root mean square error and the standard deviation of the error distribution of the battery model are linearly correlated with the corresponding SOC error statistics.As the accuracy of the battery model increases,the corresponding estimation accuracy of SOC has also been improved.Finally,after studying the basic characteristics of the battery and the impact of the battery model on SOC estimation,on the basis of the DP model,a dual unscented Kalman filter(DUKF)estimator that can simultaneously estimate the battery parameters and SOC is proposed.The results show that compared with the offline identification method,the DUKF algorithm can obtain higher model accuracy and can reflect the dynamic trend of battery model parameters in real time.The identified ohmic resistance can also be used to characterize the state of health(SOH)of the battery,so that the simultaneous on-line estimation of SOC and SOH can be achieved.Compared with the EKF and UKF algorithms,the DUKF algorithm presents an obvious advantage of accuracy,its statistical results for all SOC errors are the smallest,and the maximum error is kept within 3%.