EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC 4.0

Statistical Analysis and Regression Modelling of Lithium-Ion Battery Capacity Degradation

Changkun Luo

International Journal of Electric Power and Energy Studies · 04.11.2026 · 5 (3) · 41-51

DOI: 10.62051/ijepes.v5n3.04 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Electric Power and Energy Studies, online erschienen am 04.11.2026. Verfasst von Changkun Luo. Die Arbeit nennt 12 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

To track state of health (SOH), second-life potential and design warranties, it is necessary to accurately characterise lithium-ion battery capacity degradation. This paper compares the methodological approaches used for regression models of capacity fade, using synthetic degradation curves which are built from the structure of the publicly available Oxford Battery Degradation Dataset 1 (OBDD1) of eight 740 mAh cells discharged using an ARTEMIS urban drive-cycle at 40 °C, with each degradation profile sampled every 100 cycles. The quantitative results provided are thus illustrative, whereas the qualitative conclusions made on the problem of model selection are robust and extend to the real data. Linear, quadratic and cubic polynomials are compared to describe the capacity-versus-cycle trajectory and compared to Gaussian process regression (GPR) for prediction of end of life (EOL, 80% SOH). The quadratic model fits the dominant fade (R² ≈ 0.97), the cubic model improves the fit with the “knee” in the later life of the battery (R² ≈ 0.99), and GPR has the lowest point error (RMSE ≈ 2 mAh), but the gain over the cubic polynomial is low in absolute terms (about 0.3% of nominal capacity), and GPR's primary benefit is the well-calibrated uncertainty rather than point accuracy. From the early-life extrapolation experiment, it can be observed that the remaining useful life is consistently overestimated by low-order polynomials and thus the motivation of using uncertainty-aware approaches. The entire pipeline is made available as reproducible code, to enable direct substitution of real measurements.

Zitieren

APA

Luo, C. (2026). Statistical Analysis and Regression Modelling of Lithium-Ion Battery Capacity Degradation. International Journal of Electric Power and Energy Studies, 5(3), 41-51. https://doi.org/10.62051/ijepes.v5n3.04

Vancouver

Luo C. Statistical Analysis and Regression Modelling of Lithium-Ion Battery Capacity Degradation. International Journal of Electric Power and Energy Studies. 2026;5(3):41-51. doi:10.62051/ijepes.v5n3.04

BibTeX
@article{Luo2026,
  author = {Luo, Changkun},
  title = {Statistical Analysis and Regression Modelling of Lithium-Ion Battery Capacity Degradation},
  journal = {International Journal of Electric Power and Energy Studies},
  year = {2026},
  volume = {5},
  number = {3},
  pages = {41--51},
  doi = {10.62051/ijepes.v5n3.04}
}
Kostenlos starten →