SeqBattNet: A discrete-state physics-informed neural network with aging adaptation for battery modeling
PLOS One · 12.11.2026 · 21 (11) · e0362362
DOI: 10.1371/journal.pone.0362362 ↗Kurz gefasst
Fachartikel in der Zeitschrift PLOS One, online erschienen am 12.11.2026. Verfasst von Khoa Tran und 6 weiteren Autor:innen. Beteiligte Einrichtung: Data Science Laboratory, Faculty of Information Technology, Ton Duc Thang University, Ho Chi Minh City, Vietnam. Hauptthemen laut MEDLINE: Neural Networks, Computer, Electric Power Supplies. Die Arbeit nennt 52 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Đại học Tôn Đức Thắng.
Abstract (Originalsprache)
Accurate battery modeling is essential for reliable state estimation in modern applications, such as predicting the remaining discharge time and remaining discharge energy in battery management systems. Existing approaches face several limitations: model-based methods require a large number of physical parameters; data-driven methods rely heavily on labeled datasets; and current physics-informed neural networks (PINNs) often lack aging adaptation, still depend on many physical parameters, or require continuous state regeneration. In this work, we propose SeqBattNet , a discrete-state PINN with built-in aging adaptation for battery modeling to predict terminal voltage during the discharge process. SeqBattNet consists of two components: (i) an encoder, implemented as the proposed Hierarchical Reasoning–GRU deep learning module, which generates cycle-specific aging adaptation parameters; and (ii) a decoder, based on the equivalent circuit model (ECM) combined with deep learning, which uses these parameters together with the input current to predict voltage. The model requires only a small number of readily available physical battery parameters and, when trained on data from a single cell, demonstrates promising predictive performance on unseen battery cells within the evaluated datasets. Experimental results on three benchmark datasets (TRI, RT-Batt, and NASA) show that SeqBattNet consistently outperforms classical sequence models and recent PINN baselines. On the NASA dataset, SeqBattNet reduces the RMSE from 0.0382 to 0.0257 for battery RW4 and from 0.0406 to 0.0253 for battery RW5 compared with the state-of-the-art PDB-Net, corresponding to relative improvements of approximately 32.7% and 37.7%.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Autor:innen und Institutionen
- Data Science Laboratory, Faculty of Information Technology, Ton Duc Thang University, Ho Chi Minh City, Vietnam.
- Natural Language Processing and Knowledge Discovery Research Group, Faculty of Information Technology, Ton Duc Thang University, Ho Chi Minh City, Vietnam.
- Department of Semiconductor and Microelectronics, FPT University, Da Nang, Vietnam.
- Laboratory for Applied and Industrial Mathematics, Institute for Computational Science and Artificial Intelligence, Van Lang University, Ho Chi Minh City, Vietnam.
- Faculty of Mechanical, Electrical, and Computer Engineering, Van Lang School of Technology, Van Lang University, Ho Chi Minh City, Vietnam.
- Laboratory for Computational Mechanics, Institute for Computational Science and Artificial Intelligence, Van Lang University, Ho Chi Minh City, Vietnam.
ORCID-Profile: Khoa Tran, Hung-Cuong Trinh, Vy-Rin Nguyen, T Nguyen-Thoi, Hung Pham
Förderung
- Đại học Tôn Đức Thắng
Zitieren
APA
Tran, K., Trinh, H. C., Nguyen, V. R., Nguyen-Thoi, T., Nguyen-Thai, V., Pham, H., & Nguyen, H. S. (2026). SeqBattNet: A discrete-state physics-informed neural network with aging adaptation for battery modeling. PLOS One, 21(11), e0362362. https://doi.org/10.1371/journal.pone.0362362
Vancouver
Tran K, Trinh HC, Nguyen VR, Nguyen-Thoi T, Nguyen-Thai V, Pham H, et al. SeqBattNet: A discrete-state physics-informed neural network with aging adaptation for battery modeling. PLOS One. 2026;21(11):e0362362. doi:10.1371/journal.pone.0362362
BibTeX
@article{Tran2026,
author = {Tran, Khoa and Trinh, Hung-Cuong and Nguyen, Vy-Rin and Nguyen-Thoi, T and Nguyen-Thai, Vin and Pham, Hung and Nguyen, Ho-Si-Hung},
title = {SeqBattNet: A discrete-state physics-informed neural network with aging adaptation for battery modeling},
journal = {PLOS One},
year = {2026},
volume = {21},
number = {11},
pages = {e0362362},
doi = {10.1371/journal.pone.0362362}
}
EUVIMED