Research On Multi-state Estmation Of Lithium-ion Batteries And Sorting Method After Retirement

The development of electric vehicles is an important way to improve resource shortage and environmental pollution,and has shown explosive growth in recent years.According to the latest report of the International Energy Agency,the global sales of electric vehicles exceeded 2.1 million in 2019,a year-on-year increase of 40%.As the "heart" of electric vehicles,lithium-ion power batteries have also grown rapidly.SNE Research,an international authoritative organization about power batteries,reported that global shipments of lithium-ion batteries were 116.6 GWh in 2019,a year-on-year increase of 16.6%.Lithium-ion power batteries always deteriorate when they are used,and they usher in the first peak of retirement in 2020.According to data from China Automotive Technology Research Center,the amount of retired power batteries will be close to 25 GWh in China in 2020,a year-on-year increase of 74.4%.However,with the rapid development of electric vehicles and lithium-ion power batteries,there are many key issues that remain unresolved.On the one hand,the cell model has poor adaptability to operating conditions,resulting in low state estimation accuracy,and the calculation amount is doubled after the cells are connected in series into battery pack,which is difficult to meet practical applications;on the other hand,the research on echelon utilization of retired batteries is still in its infancy,especially the long time-consuming and high cost of sorting methods lead to extensive development,and the difficulty and low accuracy in state of health(SoH)estimation make echelon utilization risk high.This thesis takes the multi-state estimation of lithium-ion batteries and sorting method after retirement as the research content,and strives for the engineering practicability of the proposed methods on the premise of ensuring the results accuracy.The main research and innovation are as follows:Aiming at the problem of poor adaptability of battery equivalent circuit model to operating conditions,this thesis proposes an iterative identification method of model parameters.Based on complex operating conditions,the model parameters are grouped and identified,and iterated in turn until the model error meets the requirements.The obtained model is verified to have high accuracy at Urban Dynamometer Driving Schedule(UDDS),and the maximum error is less than 15 mV.Furthermore,the state of charge(SoC)is accurately estimated by the adaptive extended Kalman filter based on the above model,and the mean absolute error(MAE)and root mean square error(RMSE)are both less than 2%.Aiming at the problem that it is difficult to balance the high accuracy and low complexity in the state estimation of series-connected battery pack,this thesis proposes an estimation method for the states of series-connected battery pack based on representative cells.The law of large numbers is used to theoretically demonstrate that the state of a series-connected battery pack only depends on a few representative cells,and a selection method for representative cells is proposed using quasi-open circuit voltage and ohmic internal resistance.The states of battery pack is accurately calculated only by estimating the state parameters of the representative cells,which greatly reduces the amount of calculation.The MAE and RMSE of the SoC estimation of battery pack under UDDS working conditions are both less than 3%,and the calculation time is reduced by half compared with the traditional method,which has high practicality value.Aiming at the problem of inaccurate SoH estimation of retired batteries,this thesis makes full use of the characteristics of retired batteries with large number and wide parameter distribution,and proposes a data-driven SoH estimation method for retired batteries.The capacities of 103 LiNCM retired batteries are tested and analyzed at different temperatures and different rates.The peak coordinates of the batteries’incremental capacity curve are used as input,and the corresponding capacities are as output.SoH estimation model is trained using support vector regression,and the influence of different temperatures and rates on SoH estimation accuracy is analyzed.Finally,the estimation accuracy and sample dependence are compared with the linear regression and neural network regression.The proposed method can estimate SoH more accurately,with both MAE and RMSE less than 1.5%at room temperature,and SoH can also be accurately estimated even for limited battery samples,thereby reducing test time and improving modeling efficiency.Aiming at the low-efficiency sorting of retired batteries,this thesis proposes a sorting method based on support vector classification machine.234 LiFePO4 retired batteries are disassembled as classified samples,and a characteristic charging scheme is designed for them.The Kalman filter is used to smooth the incremental capacity curves at characteristic charging,and then the capacity characteristics and DC internal resistance are quickly extracted,which solves the problem of slow capacity feature extraction at traditional low-rate charging,and improves the sorting efficiency of retired batteries.Then the multi-classification model is trained based on the support vector machine,and the retired batteries are classified with an accuracy of 96.8%.A current detection module is designed to compare and analyze the difference of parallel currents of retired batteries before and after sorting,and some effective suggestions for the echelon utilization of retired batteries are provided.In summary,in order to solve the problem of the engineering application problems of multi-state estimation of power batteries and sorting of retired batteries,this thesis has made breakthroughs in model parameter identification for cells,SoC and peak power estimation for series-connected battery pack,SoH estimation and sorting for retired batteries respectively,and proposed a set of practical and efficient methods.The related results provide solution support for ensuring safe,efficient and reliable operation of batteries in the whole life,and they have great practical significance for solving key problems in the development of electric vehicle industry.