Research On Life Prediction Of Lithium Ion Batteries Based On Electrochemical Impedancespectral Model

Lithium-ion batteries are one of the most widely used secondary batteries,its life problems will lead to serious consequences.Therefore,the accuracy of battery life prediction has been proposed higher requirements.In this paper,based on the electrochemical impedance spectroscopy model,a new model parameters acquisition method and a combination of model and particle filter are proposed to predict the battery life.The concrete work is as follows:Firstly,the electrochemical impedance spectroscopy(EIS)model of lithium ion battery is described in detail from three scales: single particle,agglomerate and porous electrode.The paper presents the method of parameter identification in different frequency ranges to obtain the lithium-ion battery positive and negative parameters.Analysis of the sensitivity of the parameters,and finally determine the nine parameters as the high sensitivity parameters.Validate the results of parameter identification by the combination of the experimental data and the actual battery.The two kinds of verification results show that the obtained parameter error is very small and the identification result is reliable.Secondly,the battery classification method based on electrochemical impedance spectroscopy is proposed,and the battery is classified by fuzzy clustering algorithm.Then,different aging temperature and circulating current magnification under the four aging model were designed for battery cycle life test,and the battery parameters set was acquired according to different aging stage.The five parameters were analyzed by Two-Factor Analysis of Variance,and five parameters were determined as the internal health characteristics of the battery,and the regularity of the degradation of the health characteristics and the fitting results were given.Combined with the characteristics of electrochemical impedance spectrum,the aging mechanism of the battery is analyzed.Finally,the neural network algorithm are used to establish the relationship between the health characteristics parameters and capacity of the battery,and the trend of battery capacity decay can be accurately tracked.Particle filter algorithm is adopted,and the state space model is established by the battery health degradation rule of characteristic parameters,and neural network model is set up as an observer,eventually give life prediction of point estimation and the possible failure of the range.The experimental results show that the method can predict the residual cycle life of the battery.