PAHM: A Multi-Stage Aware Hybrid Framework for Lithium-Ion Battery Degradation Modeling and EOL Prediction
8th Global Power, Energy and Communication Conference, GPECOM 2026, Naples, Italy, 3 - 05 June 2026, pp.798-805, (Full Text)
- Publication Type: Conference Paper / Full Text
- Doi Number: 10.1109/gpecom70462.2026.11578600
- City: Naples
- Country: Italy
- Page Numbers: pp.798-805
- Keywords: battery degradation, data-driven modeling, EOL prediction, multi-stage aging behavior, physics-informed modeling
- Yıldız Technical University Affiliated: Yes
Abstract
Battery end-of-life (EOL) prediction remains a challenging task due to the nonlinear and multi-stage nature of degradation processes. This study proposes a hybrid framework, referred to as the Proposed Aware Hybrid Model (PAHM), which combines multi-region degradation modeling with scenario-based learning, similarity-driven estimation, and optimization-based scaling. The model is designed to capture both global degradation trends and local transition dynamics within a physically consistent structure. To improve adaptability under varying operating conditions, a dataset of 500 synthetic degradation scenarios is generated. The model is validated using partial-cycle data, where only the first 400 cycles are available to predict the remaining degradation trajectory. The results show that PAHM achieves MAE =0.0639, RMSE = 0.0784, and MAPE =1.51%, indicating competitive accuracy together with stable performance across different degradation stages. These findings suggest that the proposed approach provides a reliable solution for EOL prediction under limited data conditions.