ECA-CNN-Driven RUL Prediction for Lithium-Ion Batteries: A Charging Data Deep Learning Approach

Accurate prediction of Remaining Useful Life (RUL) for lithium-ion batteries (LIBs) is vital to ensure operating safety and optimize maintenance strategies for energy storage systems. Current challenges stem from the complex interplay of multi-scale aging mechanisms. This study presents ECA-CNN-driven framework that leverages to enhance RUL prediction accuracy. Firstly, the charging data set is preprocessed to construct four characteristic curves for voltage, current, voltage difference and current difference. Then according to different prediction needs, the corresponding input feature matrix is constructed. Finally, the feature matrix and the fused efficient channel attention-convolutional neural network model are fused to realize the RUL prediction of lithium-ion batteries. The experimental results demonstrate that the prediction accuracy of the RUL of LIBs is as high as 98.97%.