The health status estimation of power batteries based on PCA-XGBoost algorithm and multidimensional fusion features
JIANG Zhengyi
ZHAO Yaomin
WU Xueling
CHEN Siru
WANG Yanjie
HE Zenghui
Abstract：Based on the independently collected cyclic aging data of X-series(NCM811/graphite),B-series(NCM811/silicon carbon),and F-series(NCM523/graphite)lithium-ion batteries,discharge voltage and relaxation voltage data were extracted as feature sources.Principal component analysis(PCA)was applied for dimensionality reduction,and extreme gradient boosting regression algorithm(XGBoost)was used to verify the predictive performance of the proposed PCA XGBoost model in dif-ferent battery systems.The research results indicate that the cumulative contribution rates of the five principal components of X,B,and F lithium-ion batteries reached 99.88％,99.77％,and 99.61％,respectively.The use of PCA can effectively reduce information redundancy.The root mean square error(RMSE)for predicting the health status of X,B,and F lithium-ion batteries using reduced di-mensional features is only 0.0015 Ah,0.0009 Ah,and 0.0013 Ah(≤ 0.21％),indicating that the PCA-XGBoost model has good robustness and estimation accuracy while improving data efficiency when estimating the health status of cross system lithium-ion batteries.
Keywords：lithium-ion batteryprincipal component analysisextreme gradient boosting regression algorithmhealth status estimation
Publication Date：2025-06-25
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：8( 35-41,47 )