Multi-Stage State-of-Health Estimation for Lithium-Ion Batteries Based on an Adaptive SHAP-Guided Dual-Attention LSTM
4 affiliations
- 1School of Mechanical Engineering and Automation, Northeastern University, Shenyang 110819, China
- 2School of Mechanical Engineering, Ningxia Institute of Science and Technology, Shizuishan 753000, China
- 3School of Materials Science and Engineering, Northeastern University, Shenyang 110819, China
- 4Shandong Key Laboratory of Intelligent Manufacturing Technology for Advanced Power Equipment, Weifang University, Weifang 261061, China
Abstract
Accurate state-of-health (SOH) estimation across nonlinear battery aging is challenged by condition-dependent feature relevance. This study combines aging data up to a defined 80% SOH operational end-of-life threshold with an adaptive SHapley Additive exPlanations (SHAP)-guided dual-attention long short-term memory (LSTM) network. Current, voltage, state of charge (SOC), and a pulse-derived short-time resistance indicator from hybrid pulse power characterization (HPPC) are used as inputs. Training-set SHAP importance supplies a weak global feature prior, whose influence is adjusted for each input sequence by an adaptive gate. Residual feature attention recalibrates the inputs, while temporal attention aggregates representations from a shared LSTM encoder. Stage windows and labels are permuted synchronously during training to avoid fixed-position shortcuts. Under held-out SOC conditions, the complete model achieves MAE = 0.3418 percentage points (pp), RMSE = 0.5652 pp, MAPE = 0.3825%, and R2 = 0.9945, reducing MAE and RMSE by 14.1% and 12.4% relative to the Plain Shared LSTM. The results support the coordinated architecture within the tested SOC-holdout protocol; cross-cell performance is assessed separately under a different UDDS setting.
Related
- Generative Data Augmentation and Sparse-Attention Informer for Cross-Temperature SOH Estimation of Lithium-Ion Batteries
- Physics-Informed Explainable Machine Learning for Hybrid State of Health and Remaining Useful Life Estimation of Lithium-Ion Batteries
- State of Charge Estimation for Lithium-Ion Batteries Under Wide Temperature Ranges Using Error Feedforward of Sparrow-Optimized Back Propagation Neural Network
- A Physics-Informed TCN-LSTM Approach for Interpretable Internal Temperature Estimation of NMC811 Lithium-Ion Batteries
- State of Charge and State of Energy Estimation for Semi-Solid Lithium-Ion Batteries Based on Multi-Scale Temporal Convolutional Network
BibTeX
@article{li2026multistage,
title = {{Multi-Stage State-of-Health Estimation for Lithium-Ion Batteries Based on an Adaptive SHAP-Guided Dual-Attention LSTM}},
author = {Pengxiang Li and Zixi Wang and Nan Zhou and Shanshan Guo},
journal = {Batteries},
year = {2026},
month = nov,
doi = {10.3390/batteries12110458},
url = {https://www.mdpi.com/2313-0105/12/11/458},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Multi-Stage State-of-Health Estimation for Lithium-Ion Batteries Based on an Adaptive SHAP-Guided Dual-Attention LSTM
AU - Pengxiang Li
AU - Zixi Wang
AU - Nan Zhou
AU - Shanshan Guo
PY - 2026
DA - 2026/11/03
JO - Batteries
DO - 10.3390/batteries12110458
UR - https://www.mdpi.com/2313-0105/12/11/458
AB - Accurate state-of-health (SOH) estimation across nonlinear battery aging is challenged by condition-dependent feature relevance. This study combines aging data up to a defined 80% SOH operational end-of-life threshold with an adaptive SHapley Additive exPlanations (SHAP)-guided dual-attention long short-term memory (LSTM) network. Current, voltage, state of charge (SOC), and a pulse-derived short-time resistance indicator from hybrid pulse power characterization (HPPC) are used as inputs. Training-set SHAP importance supplies a weak global feature prior, whose influence is adjusted for each input sequence by an adaptive gate. Residual feature attention recalibrates the inputs, while temporal attention aggregates representations from a shared LSTM encoder. Stage windows and labels are permuted synchronously during training to avoid fixed-position shortcuts. Under held-out SOC conditions, the complete model achieves MAE = 0.3418 percentage points (pp), RMSE = 0.5652 pp, MAPE = 0.3825%, and R2 = 0.9945, reducing MAE and RMSE by 14.1% and 12.4% relative to the Plain Shared LSTM. The results support the coordinated architecture within the tested SOC-holdout protocol; cross-cell performance is assessed separately under a different UDDS setting.
SN - 2313-0105
ER -