This paper presents a data-driven approach for predicting battery capacity degradation trajectories under different storage conditions, incorporating knowledge from battery aging mechanisms and empirical data. The method is based on Support Vector Regression (SVR) with a kernel function derived from battery storage temperature, state-of-charge (SoC), and time effects. The approach is evaluated using two real datasets: Dataset A includes two storage temperatures (25°C and 45°C) and two SoCs (50% and 90%), while Dataset B includes four SoCs (20%, 50%, 70%, and 95%) under the same temperature. The results show that the SVR-based model with a knowledge-motivated kernel provides accurate predictions for future battery capacity degradation under various storage conditions. The model's flexibility and integration of knowledge improve its performance and generalization ability. The study demonstrates that the proposed approach can effectively monitor and predict battery capacity degradation, benefiting energy storage applications. The method outperforms traditional data-driven models in terms of prediction accuracy and generalization, with R² values exceeding 0.98 for most cases. The approach is validated using real battery aging experiments and shows promising results in predicting battery capacity degradation under different storage conditions. The study highlights the importance of integrating battery knowledge into data-driven models for accurate and reliable battery health management.This paper presents a data-driven approach for predicting battery capacity degradation trajectories under different storage conditions, incorporating knowledge from battery aging mechanisms and empirical data. The method is based on Support Vector Regression (SVR) with a kernel function derived from battery storage temperature, state-of-charge (SoC), and time effects. The approach is evaluated using two real datasets: Dataset A includes two storage temperatures (25°C and 45°C) and two SoCs (50% and 90%), while Dataset B includes four SoCs (20%, 50%, 70%, and 95%) under the same temperature. The results show that the SVR-based model with a knowledge-motivated kernel provides accurate predictions for future battery capacity degradation under various storage conditions. The model's flexibility and integration of knowledge improve its performance and generalization ability. The study demonstrates that the proposed approach can effectively monitor and predict battery capacity degradation, benefiting energy storage applications. The method outperforms traditional data-driven models in terms of prediction accuracy and generalization, with R² values exceeding 0.98 for most cases. The approach is validated using real battery aging experiments and shows promising results in predicting battery capacity degradation under different storage conditions. The study highlights the importance of integrating battery knowledge into data-driven models for accurate and reliable battery health management.