State-of-health estimation for the lithium-ion battery based on gradient boosting decision tree with autonomous selection of excellent features

The prediction of the health status and remaining useful life of lithium-ion batteries is very important for the safety of electric vehicles and other devices. However, due to the fact that battery residual capacity cannot be measured in real time, the estimation of battery health status is a great challenge for the management system of electric vehicles. At present, machine learning methods have been widely used in battery health state estimation. Based on the experimental data of NASA lithium-ion battery, this article proposes a model based on gradient boosting decision tree (GBDT) model framework and screens effective features from the original battery information indicators to achieve accurate evaluation of lithium-ion battery health state. In this work, many features are extracted from the original charge and discharge data of the battery, and two methods, correlation coefficient and decision tree, are used to screen initial feature, then variance inflation factor (VIF) is used for further screening, finally an efficient iterative method is used to obtain a combination of well-performing features. The validity of the residual capacity estimation method is proved by the study of NASA battery data set.

Automated summary

This article proposes a method for estimating the health status of lithium-ion batteries based on the GBDT model, utilizing machine learning techniques to accurately evaluate battery health through the selection of effective feature combinations.