SOC Estimation Of Power Battery Pack For Electric Vehicles Based On AUKF

With the rapid development of new electric vehicle technology,as one of the three key factors affecting the performance of the vehicle,the power battery has become the focus of research throughout the world.The battery management system functions mainly includes system state monitoring,balanced management,thermal management and abnormal diagnosis and protection.The state of charge(SOC)of power batteries is the most important state variables in the battery management system,which indicates the remaining battery power,and provides the basis for energy distribution management.Therefore,the accurate estimation of SOC in the electric vehicle has a very important engineering application value.First of all,this paper introduced the current development situation of the power battery,and pointed out that Li-ion battery is an ideal choice for power battery of electric vehicle by comparing the performance of all kinds of batteries;In order to have a better understanding of the performance of Li-ion battery,this paper studied the influence of the discharge rate,operating temperature and cycle times on the discharge capacity of the battery through the experiments.And based on these factors,the battery SOC is redefined.Accurate battery model is a prerequisite for accurate SOC estimation.This paper analyzed the advantages and disadvantages of several commonly used models.The two order RC equivalent circuit model which is convenient for engineering application and high precision is selected as the Li-ion battery model.After completing the feature modeling of the battery,and based on the test data of the battery,the off-line method and on-line method are adopted to identify the parameters of the model,respectively.Finally,according to the error between the output voltage of the battery model and the actual measured voltage,the validity of the battery model is verified,and the online identification method can accurately identify the parameters of the model.After the establishment of the battery dynamic model,this paper estimates the battery SOC based on UKF algorithm.Due to the shortcoming of the UKF systemnoise covariance,this paper introduces adaptive covariance matching method to estimate and correction on noise characteristics of the system in real-time,and the improved algorithm can restrain the filtering divergence and improve the precision.Through the analysis of SOC estimated results using the constant current discharge experiment and the user-defined variable current discharge experiment,it is proved that the adaptive unscented kalman filter algorithm is better than the standard unscented kalman filter algorithm,the error is less than 5% and in the presence of initial SOC error,it can quickly converge to the theoretical truth value.In the end,this paper briefly presented the difference between the SOC estimation method of the power battery pack and the estimation method of the single cell,and introduces a method of SOC estimation based on the simplified model of the battery pack.In order to verify the practical application of the proposed algorithm,the ECE experiment is carried out on the real vehicle,and then through the collection of the battery pack voltage,current and other data into the battery test equipment to get the theoretical value of SOC.Then,based on the data acquired under the actual ECE working conditions,the adaptive unscented kalman filter algorithm is used to obtain the SOC estimation value.The experimental results show the feasibility,effectiveness and accuracy of AUKF used in the power batteries pack's SOC estimation.