Research On Optimization Of Lithium-ion Battery Cycle Life Test Method Based On Extreme Learning Machine

With the increasing market demand for long cycle life lithium-ion batteries,battery manufacturers have to speed up the research and development of long life batteries.Since the cycle life test is not only time-consuming but also expensive in the battery development process,it is particularly important to optimize the test scheme.In this paper,a test optimization scheme combining physical stress acceleration and machine learning is proposed to solve the problems of long test time and low accuracy of test results existing in the existing optimization methods.The main research contents and results are as follows:(1)Analyze the influence factors of cycle life.By analyzing the mechanism of capacity decay and aging of lithium-ion batteries,the factors affecting cycle life were explored from the internal and external characteristics of lithium-ion batteries.Through comparative analysis,it is found that the temperature and current have significant effects on the cycle life of lithium-ion battery.Therefore,it is considered to enhance the temperature or current stress to accelerate the cycle life of the test process.(2)Establish the cycle life prediction model.Extreme Learning Machine(ELM)algorithm is introduced,and then a series of derived algorithms are derived.The self-encoder and hierarchical ELM based on ELM are applied to the cycle life test of lithium ion batteries.(3)Propose the optimization scheme of fusion test.A number of lithium-ion battery cycle life test standards were comprehensively analyzed and compared,and the national standard GB/T31484-2015 was selected for detailed introduction.Then,the optimization test method of increasing charge and discharge rate and the predictive test method of Extreme Learning Machine are introduced.After comparing the advantages and disadvantages of the two methods,a fusion test optimization scheme combining the two complementary methods is proposed.In order to verify the feasibility and accuracy of the proposed scheme,the TRI battery data set was selected to carry out simulation verification of the fusion test optimization scheme.The simulation results show that the relative error between the test life and the actual life is only 8.4%,and the test efficiency is improved by 66.85%.(4)Make charging and discharging equipment and verify the optimization scheme.In order to verify the actual effect of the proposed scheme,a set of lithium ion battery charging and discharging equipment was made.After the cycle experiment of the lithium-ion battery with the charge-discharge equipment,the differences in performance indexes such as time and precision between the proposed scheme and other optimized schemes were analyzed and compared.Among them,the fusion test optimization scheme took the least time,which was 43%of standard magnification and 70%of high magnification,respectively.And the root mean square error and mean absolute percentage error of the test results of this scheme are 6.6408 times and 1.1656%respectively,which are less than other optimization methods.The cycle experiment verifies the high efficiency and economy of the fusion test optimization scheme.