Life Test And State Of Health Estimation For Lithium-ion Batteries

With the worsening of energy crisis and environmental crisis,new energy vehicles have received more and more attention and development.As the power source of the current mainstream new energy vehicles,lithium-ion battery has also become a hot research topic.Accurate estimation of the health status is very important to estimate the state of charge,remaining mileage,reliable operation,and safe maintenance of lithium batteries.How to establish the battery SOH estimation model which is easy to realize in engineering and has high estimation accuracy is a key problem.In this paper,we analyzed the key factors affecting the aging of the lithium-ion battery,and a complete comparison test of accelerated aging of lithium-ion batteries was designed and implemented.Then,we established the equivalent circuit model of the lithium-ion battery and designed the health state estimation of the lithium-ion battery considering multi-dimensional aging information.Finally,we verified the health status estimation based on the aging test data of lithium-ion batteries.In this paper,the model is relatively simple to build and easy to implement in engineering.The model building depends on a large number of data from actual tests and can achieve high precision estimation.The main research contents of this paper are as follows:Based on the understanding and analysis of the aging mechanism of lithium-ion batteries,we extracted the key factors influencing the aging of the lithium-ion battery.The accelerated aging test of lithium battery was designed and completed aiming at the key factors which influence the aging,and then we accumulated sufficient actual aging test data.At the same time,the influence of temperature,current,and state of charge of the battery’s internal resistance is also fully considered in this paper.We designed the HPPC(Hybrid Pulse Power Characteristic)test at different temperatures so that the accelerated aging test is closer to the actual use situation.Based on the existing lithium-ion battery aging parameters,we combined the capacity and internal resistance parameters to characterize the degradation of lithium-ion batteries.We compared.The advantages and disadvantages of different methods which existed for estimating the health state of the lithium-ion battery.We selected the method that combining the equivalent circuit model and the neural network model was to estimate the health state of the lithium-ion battery,and the calendar aging model was used to correct the estimation results.In this paper,we compared the advantages and disadvantages of different equivalent circuit models and then chosen the DP(Dual Polarization)model to estimate the health state of the lithium-ion battery.To overcome the shortcoming that the general data-driven model can only perform open-loop estimation,we used the current and voltage collected in real-time to calculate the ohmic internal resistance and to correct the health state estimation results,which improves the estimation accuracy.The nonlinear relationship between the charge-discharge cycle curve of the battery and the health state of the lithium-ion battery was studied deeply,and a neural network model was established based on the discharge curve data to estimate the health state of the battery.A threedimensional calendar aging table of lithium-ion batteries at different temperatures and different states of charge was established to estimate the decline of battery health caused by calendar aging.The neural network model and the calendar life model are used as the supplement of the equivalent circuit model to modify the results of the health state estimation of lithium-ion batteries.Based on Matlab /Simulink software,a complete aging state estimation model of lithium-ion batteries was established and comprehensively verified by the accumulated actual lithium-ion battery accelerated aging test data.