A novel feature combination screening method for Li-ion battery state of health prediction

Abstract

Accurately predicting the state of health (SOH) of Li-ion batteries is beneficial for application management. Here, we propose an optimization scheme for feature combination (FC) to enhance the performance of the SOH prediction model. In this article, we first extract features from the partial charging voltage data of the battery for model input. Then, the Pearson correlation coefficient is combined with a machine learning model to optimize the selection of potential FCs and obtain the optimal FC. Finally, the optimal FC is applied to the SOH prediction tasks. The proposed scheme achieved significant performance improvement in multiple testing tasks and outperformed other methods, demonstrating its effectiveness and superiority.