Research On Algorithm Of Estimation Of SOC For LiFePO₄Battery In Electric Vehicle

As the issue of energy and environment becomes increasingly prominent, electricvehicle(EV) gets the attention of automobile industry in the world rapidly by the hugeadvantages of energy conservation and environment protection, so EV has got rapiddevelopment in recent years. Battery technology is one of the key factors affecting thedevelopment of electric vehicles, the management of the power battery can prolong theservice life of battery and improve the utilization efficiency of the battery pack. Batterymanagement is based on the battery state of charge(SOC), and battery SOC is used todescribe the battery remaining power, which is one of the most important parameters ofthe battery. But because the battery in usage process is influenced by the uncertainty ofvarious internal and external factors, which lead to major difficulty for the accurateestimate the battery SOC, estimate battery SOC is always a key point as well as adifficulty point in the battery technology.This article takes the lithium iron phosphate(LiFePO4) power battery as the objectof research, and mainly has done the following work: first, LiFePO4battery is taken forthe experimental study on charging and discharging. The resulting voltage, current,temperature and volume data are saved for analysis, with focus on research of thevoltage characteristics of LiFePO4battery. The OCV-SOC curve is obtained through theexperiment. Second,the advantages and disadvantages of common battery models areanalysed. This article chooses the second-order RC model, which can better reflect thevoltage characteristic. The model parameters are identified through experiment. Thesimulation is done to research the model with MATLAB, and it can be seen that thesecond-order RC model can better imitate the static and dynamic characteristics ofbattery than the Thevenin equivalent model, which is the most widely used. Afterwards,the present mainstream methods for estimation of SOC of the battery are analysed tofind their merits and demerits, especially the shortage of Kalman filtering.Aiming at thedivergence problem of using extended Kalman filter method for estimating the powerbattery SOC, due to the unknown noise characteristics, this paper adopts the adaptiveKalman filtering method to estimate battery SOC. Based on second-order RC model,simulated analysis is carried on with MATLAB. It can be seen that estimating batterySOC with adaptive Kalman filtering can get better accuracy in the case of unknownnoise characteristics comparing with Kalman filtering. Finally, the hardware platform isbuilt for real-time estimation of battery SOC. Improving the precision of estimationbattery SOC has important significance to the development of battery managementsystem, and even to the electric automobile.