Research And Realization Of State Of Charge Estimation Of Li-ion Power Battery

In order to ensure safe and reliable operating of electrical vehicles （EVs）, thebattery management system is needed to provide real-time and accurate usage statusinformation of power batteries using in EVs. Battery’s state of charge （SOC） offersendurance mileage information for the driver, directly reflects the battery’s state and isan important basis for vehicle control. Accurate estimation of SOC is a core part ofbattery management system and it becomes a hot and difficult spot of research.Lithium Iron Phosphate Power Batteries （LiFePO₄power batteries） have becomeone of the best power sources for EVs because of their perfect power performance. Inthis paper taken the LiFePO₄power battery as the research object, the followingresearches were done to estimate its SOC:1. At first the traditional SOC expression and its shortage were present, and thenthe working principle, advantages and disadvantages as well as main characteristicparameters of LiFePO₄power batteries were studied in detail. After analyzing the effectof temperature, charging and discharging rate, cycle life and self-discharging on batterymodeling and battery main characteristic parameters, considering all these factors, therevised SOC expression was got.2. The Simplification of unlimited RC networks, which represent the transientresponse in battery equivalent circuit model, was analyzed in detail. On the basis ofexperiment data, the model errors brought by one, two, three or four RC networks inseries to imitate the battery’s transient response were compared. Considering thecomplexity and precision of the battery models, the two order equivalent circuit modelwas chosen as an excellent solution. For the purpose of decreasing model error, onlineparameter identification method—Kalman filter algorithm was used to identify theparameters of battery two order equivalent circuit model.3. As the unreasonable assumptions to noise of Kalman filter method mayinfluence its estimating performance, even lead to algorithm not even slow convergence,the adaptive extended Kalman filter method was used in the battery two orderequivalent circuit model to dynamically estimate SOC. Since the mean and variance ofunknown noise was predicated and revised online, the influence of the unknown noiseto SOC estimation was reduced, and then the accuracy and feasibility of the SOCestimation method were proved by simulations. 4. The estimating performance of adaptive extended Kalman filter algorithmcombined with battery two order equivalent circuit model to estimate SOC is perfect, sothat it can satisfy engineering application requirement. On the basis of batterymanagement and monitoring system of changed pure EVs, the SOC estimatingalgorithm was verified through vehicle road test. The battery management system wasused to collect battery voltage, current, temperature and estimating SOC and themonitoring system was used to display the battery’s information as well as recordhistorical data. Through the CAN bus, the battery management system and themonitoring system can exchange information. On the basis of the structuralprogramming thought, software flow charts were designed. At last vehicle road testswere carried out to verify the practicable of the SOC estimation method.The simulation and vehicle road tests results show that the SOC estimation methodin the paper can accurately estimate SOC, which can satisfy the national standard onSOC estimation of power batteries used on EVs and has guiding significance forengineering application.