Gaussian-based data augmentation for improved prediction of axial capacity of UHPC-jacketed rectangular RC columns
Transport and Communications Science Journal · 20.12.2025 · 77 (4) · 371-384
DOI: 10.47869/tcsj.77.4.3 ↗Kurz gefasst
Fachartikel in der Zeitschrift Transport and Communications Science Journal, online erschienen am 20.12.2025. Verfasst von Hai Hoang Viet und 3 weiteren Autor:innen.
Abstract (Originalsprache)
Accurate prediction of the axial load-carrying capacity of reinforced concrete (RC) columns strengthened with ultra-high-performance concrete (UHPC) jackets is essential for reliable structural design. However, experimental data for UHPC-jacketed RC columns are scarce due to high costs and complex testing procedures, limiting the generalization of data-driven models. This study proposes a Gaussian-based data augmentation framework to enhance the predictive performance of machine learning models for estimating the axial capacity of UHPC-jacketed rectangular RC columns. An experimental database compiled from the literature is statistically analyzed, and Gaussian distribution-based techniques are employed to generate synthetic samples while preserving the statistical characteristics of the original data. Several machine learning models are developed and evaluated, with testing performed exclusively on the original experimental dataset to ensure unbiased generalization assessment. The results show that the best-performing model, CatBoost, exhibits poor generalization when trained solely on experimental data, achieving a test R² of 0.526 with MAE = 328.1 kN, MAPE = 33.5%, and RMSE = 540.9 kN. After Gaussian-based data augmentation, CatBoost performance improves substantially, reaching a test R² of 0.943, with MAE = 145.51 kN, MAPE = 10.5%, and RMSE = 222.1 kN. These results confirm that Gaussian-based data augmentation significantly enhances prediction accuracy, robustness, and generalization. The proposed framework offers a practical solution for mitigating data scarcity and supports reliable design and assessment of UHPC-strengthened RC columns.
Zitieren
APA
Viet, H. H., Dac, H. L., Bich, T. T. T., & Thanh, T. B. (2025). Gaussian-based data augmentation for improved prediction of axial capacity of UHPC-jacketed rectangular RC columns. Transport and Communications Science Journal, 77(4), 371-384. https://doi.org/10.47869/tcsj.77.4.3
Vancouver
Viet HH, Dac HL, Bich TTT, Thanh TB. Gaussian-based data augmentation for improved prediction of axial capacity of UHPC-jacketed rectangular RC columns. Transport and Communications Science Journal. 2025;77(4):371-384. doi:10.47869/tcsj.77.4.3
BibTeX
@article{Viet2025,
author = {Viet, Hai Hoang and Dac, Hien Le and Bich, Thao Tran Thi and Thanh, Tung Bui},
title = {Gaussian-based data augmentation for improved prediction of axial capacity of UHPC-jacketed rectangular RC columns},
journal = {Transport and Communications Science Journal},
year = {2025},
volume = {77},
number = {4},
pages = {371--384},
doi = {10.47869/tcsj.77.4.3}
}
EUVIMED