Document Type
Article
Publication Date
2026
DOI
10.1149/1945-7111/ae730e
Publication Title
Journal of The Electrochemical Society
Volume
173
Issue
11
Pages
110505
Abstract
This work introduces a unified interpretability-efficiency framework for lithium-ion battery state of health (SOH) prediction using hybrid deep learning architectures. We comparatively analyze four hybrid models: CNN LSTM MultiHead, CNN Feature Extractor LSTM, DNN LSTM, and DNN BiLSTM to disentangle how network topology, feature composition, and computational design influence both predictive fidelity and physical interpretability. By integrating Monte Carlo Shapley (MC Shapley), background occlusion SHAP (BoSHAP), and ablation analysis, we quantify the contribution and robustness of five electrochemical feature groups: time, capacity, voltage, dQ/dV and peaks of dQ/dV from NASA battery dataset. The results reveal a consistent dominance of differential capacity (dQ/dV) and capacity features, aligning with their electrochemical significance in phase transition dynamics and active material loss, while voltage and capacity provide stabilizing redundancy. The ablation derived Area Under the Ablation Curve (AUAC) establishes a precision-robustness tradeoff, and efficiency analyses expose an accuracy-latency Pareto frontier where DNN LSTM achieves sub 1% RMSE with compact parameterization, while CNN FE LSTM attains sub 10 ms inference suitable for embedded BMS deployment.
Rights
© 2026 The Authors.
This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 International License (CC BY 4.0) which permits unrestricted reuse of the work in any medium, provided the original work is properly cited.
Original Publication Citation
Islam, S., & Namkoong, G. (2026). Interpretable battery SOH prediction: A comparative interpretability framework for multi-architecture ML models. Journal of The Electrochemical Society, 173(11), Article 110505. https://doi.org/10.1149/1945-7111/ae730e
Repository Citation
Islam, Shafiyee and Namkoong, Gon, "Interpretable Battery SOH Prediction: A Comparative Interpretability Framework for Multi-Architecture ML Models" (2026). Electrical & Computer Engineering Faculty Publications. 602.
https://digitalcommons.odu.edu/ece_fac_pubs/602
ORCID
0009-0001-2089-1846 (Islam), 0000-0002-9795-8981 (Namkoong)