Research On Online Estimation Algorithm Of Health State Of On-board Power Battery Of Electric Vehicle

In recent years,the electric vehicle industry has been highly supported by the state and rapidly developed and occupied the market of the automobile industry,mainly because the energy it uses is electric,which has green and pollution-free characteristics and can protect the environment to the greatest extent.However,there are still some problems to be solved for electric vehicles,mainly due to the battery aging resulting in short battery life,and the timely replacement of the battery caused by safety accidents of electric vehicles,etc.Therefore,it is necessary to predict the battery life(SOH).Under the premise of timely replacement of the battery,its performance can be maximized to achieve the purpose of effective utilization of resources.In addition,due to the real-time operation of electric vehicles,how to use the data collected by the Battery Management System(BMS)to estimate the Battery SOH online has also become a top priority.In this paper,the SOH estimation of power batteries for electric vehicles is studied in depth,and a relatively accurate battery mathematical model is designed to realize the online estimation of battery life with the corresponding optimization algorithm,so as to better maintain electric vehicles and ensure their safe use.The specific work and research contents of this paper are as follows:(1)Aiming at the problem of online estimation of SOH,this paper is devoted to the estimation of SOH by using real-time observation data collected by BMS during the constant current charging process of the battery,which is completely independent of the initial state of the battery or the off-line aging prediction model established with a large amount of data.Therefore,it is more in line with the actual needs of the field of electric vehicles and easy to be used in BMS.(2)There are the following solutions for the inaccuracy of the current battery model.Firstly,on the basis of the existing model,parameters are added to make it more in line with the actual factory state of the battery.Secondly,the existing battery model is analyzed,and the battery model is optimized by introducing the SOC-OCV sub-model,which can effectively reduce the number of online identification parameters.Finally,for further optimization of the mathematical model for battery life.(3)Auto Regressive(AR)sub-models were used to characterize the internal impedance states of batteries,with the assumption that internal impedance would remain unchanged for the purpose of simplifying the model.In order to solve the problem that SOC-OCV submodel is updated frequently in battery mathematical model,a staggered time model parameter updating framework is proposed.Under this framework,the SOC-OCV model is updated regularly to reduce the computational complexity of the algorithm within a single sampling period.(4)Aiming at the problem that parameter identification in the battery mathematical model requires high real-time,accuracy and robustness of the algorithm,A fast Algorithm using nonlinear least Squares(NLS)to initialize Genetic Algorithm(GA)search range was proposed for online parameter identification.This NLS-GA algorithm can effectively solve the problem of nonlinear parameter identification,prevent the algorithm from falling into local optimal,and improve the estimation accuracy of parameter identification.