Research On Online Estimation For State Of Health Of Electric Vehicle Battery

Under the requirements of the era of sustainable development,new-energy vehicles,as a means of green and low-carbon transportation,have been promoted by various countries and societies.Among the new energy vehicles,electric vehicles account for the highest proportion.The power source of electric vehicles is the power pack,which provides pollution-free power for the vehicles in the way of cyclic charging and discharging,but with the usage of the battery,the internal aging of the battery will occur,resulting in the attenuation of the battery capacity.The attenuation degree is quantified by the state-of-health of the battery.The accurate estimation of SOH for electric vehicles is a crucial part in the use of electric vehicles,on the one hand,under the control of the battery internal management system,the wrong estimation of SOH will cause excessive charging and discharging of the battery,and affect the battery life seriously,on the other hand,accurate SOH estimation can reflect the actual use of the battery,which is conducive to the timely replacement of low-life batteries and guarantee the safety of vehicles and personnel.There are many SOH estimation methods for power batteries,but most of the scenarios are in the laboratory,which are divorced from the actual working conditions,but in reality,the measurement of vehicle SOH is mostly obtained by the tester on the spot through full charge and discharge,this method takes time and energy,and the measurement cycle is long,so it is difficult to meet the huge measurement demand.As the number of electric vehicles proliferates in the future,it is necessary to find a SOH estimation scheme that can not only meet the monitoring demand,but also combine with the actual working conditions.Based on the above situation,this thesis takes the charging data of the electric bus in operation as the object to deeply study the SOH estimation scheme based on the daily measured charging data of the charging vehicles.Firstly,this thesis introduces the source of existing charging data and preprocesses the original data,analyzes the charging voltage curve of constant current charging vehicle and its variation rule.Secondly,vehicles are generally not start charging the battery from 0,so the daily charging process is a fragment of the full charging process.Based on the existing data,this thesis verifies the analyzed law,proposes to use the inflection point to segment the charging voltage curve,and to use the historical charging information and the law of voltage curve change to complete the daily charging process.And then this thesis uses gauss process regression to fit the system state equation and completes the on-line SOH estimation with unscented kalman filter.Thirdly,the existing data were input into the model,and the model performance was evaluated from three aspects including model validity,model accuracy and model stability.This thesis draws the following conclusions: this scheme is effective for vehicle measured data estimation,the accuracy of the model can reach 3% and the accuracy is high,the fluctuation of the model estimation is less than 5% and the stability of the model is good.Finally,based on the estimation scheme,the SOH evaluation software for electric vehicle battery is designed to realize the remote downloading of data,online estimation of SOH and remote uploading of estimation results.