Application of machine learning models to predict the surface roller length of hydraulic jumps in rectangular channels
Transport and Communications Science Journal · 17.01.2026 · 77 (8) · 1024-1038
DOI: 10.47869/tcsj.77.8.1 ↗Kurz gefasst
Fachartikel in der Zeitschrift Transport and Communications Science Journal, online erschienen am 17.01.2026. Verfasst von Hùng Hồ Việt.
Abstract (Originalsprache)
The surface roller length of a hydraulic jump is an important geometric characteristic that directly influences the design of energy dissipation structures. Accurate determination of this parameter is of great practical significance in water resources engineering. Existing empirical formulas yield inconsistent results. Therefore, this study proposes and evaluates the predictive performance of four decision-tree ensemble models: Categorical Boosting (CB), Extra Trees (ET), Gradient Boosting (GB), and Extreme Gradient Boosting (XGB). Unlike previous studies, the hyperparameters were optimized using Bayesian optimization combined with repeated 10-fold cross-validation, the contribution of each input variable was quantified using SHAP values, and the stability of the results was verified through 30 random data splits. The results show that the machine learning models produce generally lower predictive errors than empirical formulas. On the representative split, CB ranked first in all four metrics (NSE = 0.984; MAE = 1.695; RMSE = 2.207; MAPE = 7.13%). Over 30 splits, CB retained the lowest mean MAE, RMSE and MAPE. Hence, CB is the most suitable predictive method, complementing traditional empirical formulas in cases where multiple hydraulic factors need to be considered simultaneously.
Zitieren
APA
Việt, H. H. (2026). Application of machine learning models to predict the surface roller length of hydraulic jumps in rectangular channels. Transport and Communications Science Journal, 77(8), 1024-1038. https://doi.org/10.47869/tcsj.77.8.1
Vancouver
Việt HH. Application of machine learning models to predict the surface roller length of hydraulic jumps in rectangular channels. Transport and Communications Science Journal. 2026;77(8):1024-1038. doi:10.47869/tcsj.77.8.1
BibTeX
@article{Vit2026,
author = {Việt, Hùng Hồ},
title = {Application of machine learning models to predict the surface roller length of hydraulic jumps in rectangular channels},
journal = {Transport and Communications Science Journal},
year = {2026},
volume = {77},
number = {8},
pages = {1024--1038},
doi = {10.47869/tcsj.77.8.1}
}
EUVIMED