Research On Model And State Of Charge Estimation For Lithium-ion Battery In Electric Vehicles

The increasingly serious energy crisis and environmental issuespromote the development of electric vehicles(EV)all over the world.The battery and battery management system(BMS)are keysto electric vehicles.Lithium-ion batteries have been widely used because of its no pollution,high specific energy,no memory effect and other advantages.The state of charge(SOC)estimation is the basis of BMS,which is of great significance to prolong the service life of the battery and improve the performance of the vehicle.This paper focuses onlithium-ion battery and its SOC estimation methodsbased onequivalent circuit model.Sliding mode observer(SMO)method,radial basis function(RBF)method and advanced vehicle simulator(ADVISOR)method are presented to estimate and test SOC.The main work includes:Firstly,SOC estimation methods and the equivalent circuit models are summarized and compared.The principles and performances of lithium-ion battery are analyzed.RC equivalent circuit model is established for this paper.According to thepulse current discharge(PCD)test,the proposed model is verified and the model parameters are obtained.Results show that the equivalent model used in this paper can well simulate the characteristics of battery and has higher precision.Secondly,sliding mode observermethod is proposed to study SOC estimationbased on RC equivalent circuit model,a method to improve the chattering problem versus traditional SMO.The design of sliding mode observer has a new switching function instead of sign function in the traditional sliding mode observer.The stability isprovedby Lyapunov theory and the results prove that the advantages in improved sliding mode observer which weakens the chattering and has a faster convergence rate and good tracking performance.Thirdly,a method based on RBF neural network observer for SOC estimation is put forward.The voltage estimation in the model is used as the inputparameter of neural network observer,the estimated value of disturbance is output,and then the SOC estimated value is calculated.The Lyapunov theory is used to design the neural network adaptive weightvector law and prove the stability of the designed observer.The simulation results show that the proposed method is effective and feasible.What is more,it has good effectsforSOC estimation and external disturbance approximation.Finally,the ADVISOR simulation environment is applied to build EV model and two kinds of actual working road conditionsare selected.SOC estimation simulation experiment is tested based on battery data provided by ADVISOR and algorithms areverified in the actual driving conditions.The results show that under different driving conditions,two methods can be used to estimate the SOC effectively and the estimation error is stable within 3%comparing with the reference value.