EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

A State of Health Estimation Method for Lithium-Ion Battery Packs Using Two-Level Hierarchical Features and TCN–Transformer–SE

Chaolong Zhang, Panfen Yin, Kaixin Cheng, Yupeng Wu, Min Xie, Guoqing Hua, Anxiang Wang, Kui Shao

Batteries · 01.04.2026 · 12 (4) · 123

DOI: 10.3390/batteries12040123 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Batteries, online erschienen am 01.04.2026. Verfasst von Chaolong Zhang und 7 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 32 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: major project of basic science (natural science) research in colleges and universities of Jiangsu Province, “Qinglan Project” for universities in Jiangsu Province, scientific research foundation for high-level personnel in Jinling Institute of Technology.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This study proposes a novel state of health (SOH) estimation method by extracting two-level hierarchical features linked to fundamental degradation mechanisms. At the module level, the length of the incremental power curve during constant current charging is extracted, capturing cumulative effects of subtle changes. At the cell level, a combined temperature-weighted voltage inconsistency curve is constructed. The state of charge (SOC) at its distinct knee point within the high-SOC range is a key indicator, signifying the accelerated failure stage where polarization and thermoelectric feedback intensify. This knee-point SOC quantitatively reflects the degree of SOH degradation, making it a valid feature for accurate SOH estimation. The proposed Temporal Convolutional Network–Transformer–Squeeze-and-Excitation (TCN–Transformer–SE) model assigns weights to these features via Squeeze-and-Excitation (SE) and uses Temporal Convolutional Network (TCN) and Transformer branches for parallel local and global temporal decisions. Aging experiments demonstrate the method’s superiority through multi-feature comparison, ablation studies, and benchmark evaluation, achieving a maximum mean absolute error (MAE) of 0.0031, a root mean square error (RMSE) of 0.0038, a coefficient of determination (R2) of 0.9937 and a mean absolute percentage error (MAPE) of 0.3820. The work provides a fusion estimation framework with enhanced interpretability grounded in electrochemical analysis.

Förderung

  • major project of basic science (natural science) research in colleges and universities of Jiangsu Province
  • “Qinglan Project” for universities in Jiangsu Province
  • scientific research foundation for high-level personnel in Jinling Institute of Technology
  • college students’ innovative entrepreneurial training plan program

Zitieren

APA

Zhang, C., Yin, P., Cheng, K., Wu, Y., Xie, M., Hua, G., Wang, A., & Shao, K. (2026). A State of Health Estimation Method for Lithium-Ion Battery Packs Using Two-Level Hierarchical Features and TCN–Transformer–SE. Batteries, 12(4), 123. https://doi.org/10.3390/batteries12040123

Vancouver

Zhang C, Yin P, Cheng K, Wu Y, Xie M, Hua G, et al. A State of Health Estimation Method for Lithium-Ion Battery Packs Using Two-Level Hierarchical Features and TCN–Transformer–SE. Batteries. 2026;12(4):123. doi:10.3390/batteries12040123

BibTeX
@article{Zhang2026,
  author = {Zhang, Chaolong and Yin, Panfen and Cheng, Kaixin and Wu, Yupeng and Xie, Min and Hua, Guoqing and Wang, Anxiang and Shao, Kui},
  title = {A State of Health Estimation Method for Lithium-Ion Battery Packs Using Two-Level Hierarchical Features and TCN–Transformer–SE},
  journal = {Batteries},
  year = {2026},
  volume = {12},
  number = {4},
  pages = {123},
  doi = {10.3390/batteries12040123}
}
Kostenlos starten →