Research On Full Life-cycle State Estimation Of Lithium-ion Batteries And Sorting Method After Retirement

To solve the energy crisis and environmental pollution problems,new energy vehicles instead of traditional fuel cars are developing rapidly,and by the end of 2022,national new energy vehicle fleet reaches 13.1 million units.Lithium-ion batteries have emerged as the mainstay of automotive power batteries because of their high energy density,long cycle life and other advantages.Meanwhile,statistics from the China Automotive Technology Research Center show that the accumulated retired batteries in China reached 38.02 GWh by 2021,and the retired volume is expected to reach 137.4 GWh in 2025.With the use of electric vehicles,the performance of lithium-ion power batteries will inevitably degrade.Therefore,behind the rapid development of new energy vehicles there are many problems that need to be solved.On the one hand,the current state prediction of power battery stays at the stage of vehicle application,and there is less research on the state prediction of retired batteries,and the predicted state is not updated in real time and the accuracy is low;on the other hand,the research on the secondary utilization of retired batteries is in the initial stage,and the key challenge is to improve the sorting efficiency while ensuring the accuracy in the face of the sorting work of large batch of batteries.To address the above challenges,this paper focuses on the whole life cycle from the vehicle application stage to the retirement stage,and takes the lithium-ion power battery state of health(SOH)estimation,thermal behavior analysis and retirement sorting method as the research,and the main research contents are as follows.This paper focuses on ternary lithium-ion batteries(LiNCM)as the research object.In this paper,a precision instrumentation test platform is built and different test protocols are designed for the research object.Different aging data are obtained for the whole life cycle of the battery,and the aging characteristics are analyzed to provide the necessary data and theoretical basis for the subsequent research.The traditional SOH estimation methods for Li-ion batteries mainly focus on the on-board application phase and rarely take into account the decommissioning phase.To address the problems of poor generalization ability,low estimation accuracy and long time consumption of traditional methods,a neural network SOH estimation model based on firefly optimization algorithm(GSO)is proposed in this paper.First,the incremental capacity(IC)curves of the battery under different aging are obtained during the battery single cell life cycle test.Then,the aging feature points are extracted from the IC curves.The aging feature points are used as the input of the SOH estimation model,and the real SOH is used as the output to train the model.The model has a high accuracy for SOH estimation under the whole life cycle,with the average relative error less than 1.35%and the root mean square error less than 3.3%.Also.The feature points used in this paper only require partial discharge test data,thus reducing the measurement time.With the frequent occurrence of spontaneous combustion accidents in electric vehicles as well as energy storage power plants,the thermal safety of batteries is receiving more and more attention.The available capacity of power batteries gradually decreases during actual use until they are retired,and the capacity consistency of large batch of retired batteries is poor,and there is a lack of effective means to quickly obtain accurate nuclear temperature data of large batch of different aging batteries.To address the above problems,this paper proposes a method to analyze the thermal behavior of lithium-ion batteries based on the whole life cycle electrothermal coupling model.First,the second-order equivalent circuit is used as the electrical model to calculate the characteristic test data during the life cycle test and obtain the parameter identification results of the electrical model under different aging.Then,we finish the identification of the parameters of entropy thermal coefficient under different aging,consult the measurement to obtain the specific heat capacity,thermal diffusion coefficient and other parameters of the battery,and construct the thermal model according to Bernardi’s heat generation rate formula,heat dissipation formula,energy conservation law and Fourier’s law.The outputs of the electrical and thermal models are used as the inputs of each other,which in turn affect the respective outputs,thus constructing the coupled electrical and thermal models.The maximum absolute error of the model under different operating conditions is 0.81℃ and the maximum relative error is 2.81%.Finally,to address the problems of large differences in the initial state of large-scale retired lithium-ion batteries,low sorting efficiency,and difficulty in considering the core temperature of batteries,this paper proposes a fast sorting method for retired lithium-ion batteries based on semi-supervised fuzzy C-mean clustering algorithm(SSFCM),which integrates ICMAX,discharge power and temperature aging characteristic points.In this paper,104 samples are selected for fast sorting of retired batteries.Firstly,the SOH estimation value of the battery is obtained based on the IC curve characteristic points,and the core temperature characteristic points are obtained using the electrothermal coupling model.Delineate the SOH demarcation line,and do not do classification markings for the battery singletons in the demarcation line area that are more dense.The cluster center is initialized using the marked samples,and then the influence ratio of ICMAX,discharge power and core temperature characteristic points is integrated to complete the fast sorting of retired Li-ion batteries based on SSFCM algorithm.The model binning accuracy is 96.15%.