Electric Vehicle Battery State-of-Charge Estimation Based On Adaptive Kalman Filtering

With the continuous development of new energy vehicles, power battery, whichinfluences the performance of vehicle as one of the key factors, has become a hot spotaround the world. For the power battery, monitoring states, energy management andsafety control are important segments to improve the performance of power batteryand reduce the costs of maintenance. Among all the management factors of powerbattery, the battery’s State of Charge (SOC) is one of the most important one, and itrepresents the remaining power of the battery, which is an important parameter of theother functions of battery management, and provides the basis for the energydistribution at the meantime.In the study of this paper the lithium ion battery was the research object, themain work including:Firstly, expounding the significance of power battery’s SOC estimation,introducing the concept of battery management system, the definition of SOC and thefactors which influence SOC respectively, and analyzing the current status of batterySOC estimation methods and the battery models.Then the working principles of lithium ion batteries are summarized, and themathematical model which based on the electrochemical principles is established,next the equivalent circuit model is deduced from mathematical model. Based on theequivalent circuit model established, the state space model of the battery wasestablished. To obtain the parameters of the battery model, experiments whichcalibrate the relationship between the battery open circuit voltage and SOC and whichidentify the battery’s resistance and RC circuit parameters are carried out. In order toverify the accuracy of the model, the output of the battery model which parametershad been determined and the actual output of battery are compared.Based on the theory of the Kalman filter, an adaptive Kalman filter method isproposed to estimate the lithium ion battery’s SOC. Since the adaptive Kalman filteralgorithm can estimate the mean and variance of unknown noise online, the influenceof the unknown noise is reduced. Based on adaptive Kalman filter, the SOCestimation model is established in Simulink. To verify the accuracy of the battery’sSOC estimated by adaptive Kalman filter, and compare with Kalman filter, simulationexperiments by using the battery model in Simulink is conducted.Finally, the AMESim and Simulink joint simulation experiments are conducted. In the environment of AMESim vehicle model, the operation data of AMESim vehiclemodel are imported to the S-Function Interface in Simulink through an interface ofAMESim. Combined with the adaptive Kalman filter estimation algorithm establishedin Simulink, joint simulation experiments, which are under different operationconditions, are carried out for the dynamic simulation model. Then comparing withKalman filter, the advantage of the adaptive Kalman filter is verified, and thesimulation results are analyzed at the end.