Health State Estimation And Abnormal Condition Analysis Of Lithium-ion Battery Based On Voltage Characteristics

Recently the share of electric vehicles in the automotive market has exploded.The increase in the number of electric vehicles has put forward higher requirements for the cycle life and reliability of power batteries.It is of great significance to regularly estimate the battery state of health(SOH)and research the rapid capacity decay caused by abnormal working conditions.In terms of aging estimation,modelbased aging estimation involves relatively too complicated parameter identification,and it is difficult to apply in real vehicles.In terms of battery capacity abnormal attenuation and troubleshooting,ex-situ analysis involving chemical mechanisms requires to destroy and disassemble the battery and cannot be applied in real vehicles.In response to the above problems,two cylindrical lithium-ion batteries are used as the research object to conduct experiments.The capacity increment curve and the differential voltage curve are obtained through the voltage and capacity changes during the charging process,so as to realize the estimation of the battery aging state and research on abnormal decay of battery capacity.The main research contents are as follows。Carry out cycle aging experiment and aging analysis on the batteries,to obtain the capacity increment curve and the differential voltage curve,and the influence of different factors on the capacity increment curve is researched.The law of the two curves with the battery cycle number is analyzed,and the aging mechanism is combined to identify the capacity decay mode of the battery.Gaussian function is used to fit the different peaks of the capacity increment curve under different cycles,and to characterize battery aging through the change of function parameters.The Pearson correlation coefficient method is used to correlate the characteristic parameters,to determine the model input parameters for SOH estimation.Taking the above characteristic parameters as the model input and the battery’s actual SOH as the model target output,the battery SOH is estimated and the accuracy is compared and verified by using multiple linear regression,random forest method and BP(Back Propagation)neural network.The results prove that above methods can achieve good estimation results,which verifies the effectiveness of the characteristic parameter.The overall estimation effect of the BP neural network is the best.Carry out abnormal working condition experiment and result analysis on the batteries.Abnormal conditions experiments include overcharging-overdischarging,high-rate charging and low temperature charging-discharging.The capacity attenuation and capacity increase curve of the overcharging-overdischarging and high-rate charging test battery are compared with the normal aging battery.Through the abnormal change characteristics of the curve peak,basic identification and distinction can be realized between the overcharging-overdischarging and high-rate charging conditions.Differential voltage curve of the battery under low-temperature charging-discharging can be obtained and the battery’s lithium deposition in low temperature can be characterized through the abnormal peaks appearing in the curve.Through the above research,the capacity increment curve and differential voltage curve can be obtained by using the voltage and capacity changes during the charging and discharging process of the battery,and then perform accurate estimation of the battery SOH and basic identification of abnormal battery conditions.The abnormal capacity attenuation of the battery can be detected and diagnosed,to improve the service life of the battery.