Research On SOH Estimation Method For Pure Electric Vehicle Li-ion Traction Battery

Facing the desperate resource shortage and deteriorating environmental pollution,the development and research of pure electric vehicles with lithium-ion battery as the main power source has gradually attracted the close attention of automobile manufacturers.Accurately real-time estimating the SOH(State of Health)of the lithium-ion battery has great significance in terms of providing the safe operation of the automotive electrical system,extending the rest useful life of the traction battery,and increasing the cruising range of the electirc vehicles.Based on the data in cycle life test of lithium-ion batteries provided by the NASA PCoE Research Center,serveral health factors(Health Indicator,HI)are extracted by the correlation analysis method.And then the polynomial regression model,which chooses the health factors as the independent variable to estimate the SOH of batteries,is constructed and fitted.Firstly,in the earlier stage of the constant current charging process of the lithium-ion battery,the interval between constant range of voltage rise curve(the interval when the terminal voltage of battery is located in 3.90V?3.95 V,3.95V?4.00 V and 4.00V?4.05V)are extracted as the health indicator to estimate the SOH.When just using quadratic fit,the root mean square error RMSE of polynomial regression is limited in 0.04,the model fitness coefficient R2 is higher than 0.98,and the relative error is less than ± 2%.Secondly,in the medium term of the constant voltage charging process of the lithium-ion battery,the interval between constant range of current drop curve(the time it takes for the charging current to fall from 0.75 A to 0.25 A,and from 0.50 to 0.25A)are extracted as the health factor to estimate SOH.When just using quadratic fit and training samples whose true SOH is higher than 75%,the RMSE of polynomial regression is limited in 0.042,the R2 is higher than 0.95,and the relative error is less than ± 3%.Finally,during the charging process of the lithium-ion battery,the interval between constant range of temperature drop curve(the time it takes for the temperature of battery to descend from the highest point and the falling range of the temperature is 0.5??1.5?and 1.0??2.0?)are extracted as the health indicator to estimate the SOH.When just using cubic fit and training samples whose true SOH is lower than 87%,the RMSE of polynomial regression is limited in 0.04,the R2 is higher than 0.97,and the relative error is less than ± 3%.Based on the ageing sample data of lithium-ion battery provided by the NASA PCoE Research Center,robust health factors are extracted and concise polynomial regression model for estimating SOH are built.The fitting results indicate that this method for predicting SOH not only own obvious advantages of clear parameter measurement and strong adaptable algorithm,but also can accurately estimate the SOH of the lithium-ion traction battery.Thus,this thesis certainly has reference value and practical significance.