Study On The Health Feature Extraction And Diagnosis Of Power Lithium-Ion Batteries

Lithium-ion batteries play an important role as the key part of energy supply of electric vehicles(EVs).The lithium-ion batteries,state of health(SOH)is a comprehensive evaluation index of battery aging,and characterization and estimation of batteries' SOH is the key technology of the next generation of battery management system.Based on laboratory test data,the paper studied three core issues,which are SOH feature extraction,SOH estimation and battery fault alert.The detailed research and achievement are as follows:Firstly,different cycle life experiments were designed for the two types of batteries to analyze the aging decline characteristics of batteries under different environment temperature and different rates.With the actual capacity as the main evaluation parameter,the effect of ambient temperature and current rates on the aging of the battery is preliminarily obtained by comparing and analyzing the experimental data.In order to study the batteries,cycle characteristics under working conditions,the paper presented a working conditions extraction method based on the equivalence of power and power variation.Based on the assumption of power equivalence only in the general condition extraction method,the equivalence of the change of power which reflects the power mutation capacity of the battery is added to the target function,and the equivalence between the target operating condition and the original operating condition is further enhanced.Secondly,aimed at batteries from Manufacturer I which present linear aging characteristics,using incremental capacity analysis method,the variation characteristics of NCM batteries' incremental capacity curve(IC curve)with the aging of batteries are studied,and the characteristic parameters of the IC curves closely related to the battery ageing mechanisms are analyzed.The capacity estimation model of NCM lithium-ion battery to indicate batteries' SOH is established.To resolve the serious multicollinearity among the character parameters,the capacity estimation model based on the principal component regression is proposed.The experimental results demonstrate that the estimated errors are less than 2%.Using the parameters reflecting the electrochemical characteristics of the batteries,the proposed method not only estimate SOH of the battery,but also can identify the aging mode of the battery,providing the foundation for battery life management strategy.Thirdly,aimed at batteries from Manufacturer ? which present nonlinear aging characteristics,the change characteristics of the IC curve of the low current charging data are studied,and the quantitative calculation of the thermodynamic loss of the batteries is realized.The entropy weight method is used to select and simplify the characteristic parameters of the IC curve,and the support vector machine is used to establish the regression model between the characteristic parameters and capacity.Simulating the actual use of data acquisition and model training methods for model training,the results show that the test error is less than 1.5%,showing that the model has good estimation precision.Finally,in order to solve the malfunction such as drum that may appear during the use of non-linear degenerate batteries,a battery fault warning model is established.The quantile regression method that can obtain multiple regression lines at the same time is used to model,and the normality of the regression residuals is verified.A band-shaped safety zone based on the nature of the normal distribution is established,and the model alarms when the capacity beyond the region.Then the model is compared with other similar modeling methods,which shows the advantages of the modeling method.