Research On SOC Estimation For Lithium Battery Of Electric Vehicle

A critical component of battery management system and electic vehicle system is the accurateand real time knowledge of SOC when running on road. Some theories are researched in this paperaiming at the estimation of state of charge. Several high efficient and accurate estimation algorithmsare proposed, and some available conclusions are summarized. In this thesis, as LiFePO4battery beingfor research object, SOC estimation methods of lithium-ion battery are studied. The specific resultsare as follows:(1) From the basic principles of lithium-ion battery, based on the related tests, the voltage feature,resistance feature, efficiency feature and circulation feature of lithium-ion battery are discussed.Taking the direction of charge and discharge, charge and discharge rate and temperature factors intoaccount, the electrochemical composite model of battery is established, the parameters of the batterymodel are obtained using least-square method. The simulation model is established inMATLAB/SIMULINK software, simulation results show that the battery model has high precisionand can simulate the dynamic characteristics of battery accurately.(2) The particle filter is applied to the lithium battery SOC estimate. Using the effective particlenumber and system resampling technology to estimate the SOC of lithium battery based on theanalyzing the basic principle and existing problems of particle filter algorithm, particle degradationphenomenon and resampling technology. The results of virtual experiment based on ADVISORdemonstrate that PF is available in SOC estimation.(3) In order to better solve the particle degradation problem and guarantee the diversity ofparticles, the particle filter algorithm is improved in this paper. Using the EKF to generate theimportant density function on the basis of the resampling method, design the calculation process ofSOC estimation, the results of virtual experiment based on MATLAB demonstrate that EKPF isavailable.(4) The unscended kalman particle filter is applied to the lithium battery SOC estimate. EKFmust calculate the Jacobi matrix of the nonlinear fuction, prone to errors in the complex model, andintroduce the linear error when linearization. But UKF avoids the tedious Jacobi matrix calculation,with high precision. So can use the UKF with symmetric sampling strategy and scale correction togenerate the important density function based on paticle filter algorithm, further improve theestimation precision. The simulation results show that the UPF algorithm has unique superiority when estimate the battery SOC.