Battery SOC Estimation Based On High-end PNGV Model

As the definition of power battery residual capacity, SOC(state-of-charge)can not only visually reflects the performance of the battery pack such as the available capacity of the battery pack, consistency,but also an important prediction of he dynamic performance of electric vehicles,like mileage basis and Gradeability. In order to optimize the energy use of the battery pack and guarantee the safety of EV,Accurate SOC estimation has been considered as the key technology of the Battery Management System. In this dissertation, the battery SOC estimation based on lithium-ion batteries is studied,And the work can be mainly divided into two aspects as the following:First part is research on Battery model.As one important precondition of obtaining accurate SOC estimation,An accurate model of the battery should be established.Firstly,basic characteristics of the battery is studied and analyzed through characterization Experiments.Considering requirements of the battery model applied in the actual Battery Management System and basic characteristics of the battery,Second-order PNGV battery model is established.Then parameters of this model includes open circuit voltage (Uoc) and RC parameters are identified according to the principle of least squares. The accuracy of the battery model is verified at last.The second part of research in this dissertation is estimation algorithms. Two Nonlinear Kalman filtering algorithms are introduced to estimate the SOC of battery to deal with the nonlinearity of the improved PNGV model.The results of simulation shows that UKF algorithms do improve the SOC estimation accuracy at the cost of lowering estimation efficiency.To solve this problem, A new unscented Kalman filter algorithm based on RB decomposition (RBUKF) is proposed and implemented to estimate the SOC of the battery. Finally,both accuracy and efficiency of this new estimation algorithm are demonstrated by simulation results.