Kernel-driven prediction of discharge coefficient in semi-circular side weirs: a comparative machine learning perspective

Articles in Press

Document Type : Research Paper

Authors

1 Department of Water Engineering, Faculty of Civil Engineering, University of Tabriz, Tabriz, Iran

2 Department of Water Resources Engineering, Faculty of Civil Engineering, University of Tabriz, Tabriz, Iran

Abstract

Accurate prediction of the discharge coefficient in sharp-crested semi-circular side weirs is of high importance due to its critical role in hydraulic analyses, optimal flow management, and the safe design of water structures. In this study, to enhance the prediction accuracy of this coefficient, the Support Vector Machine (SVM) was integrated with two metaheuristic optimization algorithms: the Horse Optimization Algorithm (HOA) and the Reptile Search Algorithm (RSA). Initially, the dimensionless parameters influencing the discharge coefficient were identified, and several predictive models were developed. The experimental dataset was randomly split into training (80 percent) and testing (20 percent) subsets. Results demonstrated that both models exhibited high predictive accuracy; however, the SVM-HOA model delivered superior performance during the testing phase (NSE = 0.887, RMSE = 0.025, R2 = 0.899), yielding more precise outcomes. Sensitivity analysis further revealed that the ratio of upstream flow depth over the weir crest to the weir diameter (h1/D) and the ratio of weir crest height to channel width (P/B) were the most influential parameters in modeling the discharge coefficient. Overall, the findings confirm that the proposed SVM-HOA hybrid model provides a powerful tool for accurately predicting the hydraulic behavior of semi-circular side weirs under various operational conditions.

Keywords

Main Subjects

Iranian Journal of Soil and Water Research

Articles in Press, Accepted Manuscript
Available Online from 06 October 2025

Files

Share

How to cite

Statistics