Accurate prediction of lithium-ion battery lifespan is vital for ensuring operational reliability and reducing maintenance costs in applications like electric vehicles and smart grids. This study introduces a novel hybrid learning framework that addresses the limitations of existing methods by integrating multisource data fusion with a stacked ensemble (SE) modeling approach, specifically designed to handle the heterogeneity and variability in battery data from diverse sources. By leveraging heterogeneous data sets from the National Aeronautics and Space Administration (NASA), Center for Advanced Life Cycle Engineering (CALCE), MIT-Stanford-Toyota Research Institute (TRC), and nickel cobalt aluminum (NCA) chemistries, a variance-aware weighting mechanism mitigates variability across data sets. The SE model combines ridge regression, long short-term memory (LSTM) networks, and eXtreme Gradient Boosting (XGBoost), effectively capturing temporal dependencies and nonlinear degradation patterns. The proposed framework not only achieves superior predictive performance, with a mean absolute error (MAE) of 0. 0058, root-mean-square error (RMSE) of 0. 0092, and coefficient of determination (R2) of 0. 9839, but also demonstrates a significant improvement over baseline models, including a 46. 2% increase in coefficient of determination and an 83. 2% reduction in root-mean-square error. Shapley additive explanations (SHAP) analysis identifies differential discharge capacity (Qdlin) and temperature of measurement (Tempₘ) as critical aging indicators. This scalable, interpretable framework enhances battery health management, supporting optimized maintenance and safety across diverse energy storage systems.
He et al. (Fri,) studied this question.