Modeling And SOC Estimation Of Ternary Lithium-ion Batteries For Electric Vehicles

In recent years,the development of electric vehicles has received the attention of all countries with the aggravation of global environmental and energy problems.As the most important energy storage component of electric vehicles,lithium-ion batteries still have a series of problems,which restrict the further development of electric vehicles.In order to utilize the limited power of electric vehicles more effectively,battery management system(BMS)should estimate the state of charge(SOC)of batteries accurately.Battery model which is a centralized mathematical expression of the external characteristcs of lithium-ion battery,plays an important role in many SOC estimation algorithms.In this paper,the most widely used ternary lithium-ion batteries in electric vehicles are taken as the research object,and their modeling scheme and SOC estimation are studied.Starting with the working principle of lithium-ion batteries,this paper introduces the structure and composition of lithium-ion batteries and analyses the main chemical reaction principles inside them.Three kinds of equivalment circuit models which are widely used at present are analyzed and compared,and the second-order Thevenin model with high accuracy and simple structure is selected to describe the external characteristics of lithium-ion battery finally.Experiment is designed to study the relationship between battery's parameters and their influencing factors,and the influence of fitting polynomial number on the accuracy of OCV-SOC fitting curve is analyzed.Electric vehicles can make a safe and efficient battery use strategy according to the research of battery characteristcs.After analysing the principle and process of identifying the second-order Thevenin model by Recursive Least Square(RLS)algorithm,this paper found that the model built by RLS algorithm sometimes can not guarantee the effectiveness of both RC segments,and the model has some limitations in accuracy and adaptability.In order to solve this problem,this paper designs a new type working condition for parameter identification,and a recursive least square algorithm with constraint factors based on this condition is proposed for battery modeling.This modeling scheme refers two different working conditions at the same time,so the two RC segments in the battery model are valid,and it is proved that the model built by this modeling scheme can express the voltage change of battery in most current and temperature conditions and has a high accuracy and adaptability.In the part of SOC estimation,this paper analyses the error sources of the amperehour integration method,and focuses on the process of SOC estimation using Extended Kalman Filter(EKF)which is based on battery model.Experiment results show that EKF algorithm has a high accuracy in most cases,and compared with the ampere-hour integration method,EKF algorithm has stronger robustness in SOC estimation.However,when the discharge depth reaches 10%,the error of SOC estimated by EKF algorithm increases significantly due to the reduction of the accuracy of battery model.To solve this problem,this paper designs an "EKF-AH" joint SOC estimation scheme,which is still based on EKF algorithm,but the "EKF-AH" joint SOC estimation scheme relies more on the ampere-hour integration method for SOC estimation when battery is in low power.It has been proved that the "EKF-AH" joint SOC estimation scheme has higher SOC estimation accuracy than EKF algorithm when the battery is in low power.Finally,a battery management system which can manage most 12 batteries based on MC33771 was designed in this paper.The system uses TI's TMS28377 D chip as the main control unit,and MC33771 designed by NXP is chosen as the battery management chip which uses SPI to communicate with the main control unit.The additional resistance between sampling line and battery is identified by pulse current method,which eliminates the deviation between sampling voltage and actual battery voltage.On-line SOC estimation results show that the joint SOC estimation scheme has high accuracy in battery management system and has a certain practical prospect.