Research On Soc Estimation Of Lithium Battery Used In Electric Vehicles Based On Strong Tracking Kalman Filter

As the problems of energy shortage and environment pollution arise, new energy electric vehicles gradually attract people’s attention. The electric vehicle battery management system, as an important component of electric vehicles, has become a popular research subject for many auto makers and scientists. State of charge is a key performance index of efficient management for electric vehicle battery management system. The accurate estimation of SOC directly affects not only the battery management system performance but also the vehicle commercialization. However, it is difficult to carry out accurate estimation of SOC because of the battery nonlinear dynamic performance in different statuses and its complexed reaction process. Research on SOC is still a new subject abroad. Thus, it is significant to explore the accurate estimation method for SOC to develop electric vehicles.This paper completes battery modeling, parameter identification and SOC estimation by Matlab, where lithium battery is chosen to be research object. All the specific work is shown as follows:First of all, the paper analyzes the influence of discharge rates, temperatures and cycle times on battery capacity, and finds out the compensation relations of discharge rates on lithium battery capacity. It also studies the characteristics of springback voltage and open circuit voltage by carrying out pulse discharge and intermittent experiments, which helpfully conducts the following lithium-ion battery modeling based on the voltage response curve.Secondly, by comparing the electrochemical model, neural network model and equivalent circuit model, the second-order RC model is chosen as the battery equivalent circuit model, and combining the ampere-hour integral equation to establish the state space model of the battery. Considering the time-varying characteristics of the model parameters, through HPPC test data, the exponential fitting method is adopted to identify impedance parameters in different SOC. Then, the paper establishes battery model in Matlab/Simulink and verifies the accuracy of battery model and identification parameters.Finally, due to the nonlinear characteristics of battery system, this paper applies the extended kalman filter which can be used in nonlinear system state observation to estimate SOC, and then compares its performance with that of traditional ampere-hour integral algorithm. The simulation results show that the extended kalman filter performs better than traditional algorithm in eliminating the accumulated errors and correcting initial errors. Therefore, the paper puts forward a strong tracking kalman filter algorithm in that extended kalman filter relies heavily on the accuracy of battery model and has the poor tracking mutation ability. Based on the extended kalman filter algorithm, by introducing a suboptimal fading factor, it forces the output residual sequence to be orthogonal, and improves the traceability of this algorithm when the state changes or the model is uncertain. The results of simulation experiments on the custom pulse charging and discharging show that this algorithm has faster convergence speed and better accuracy than the extended kalman filter.