Sediment transport modeling in non-deposition with clean bed condition using different tree-based algorithms.

Area of Science:

• Hydraulic Engineering
• Computational Fluid Dynamics
• Environmental Engineering

Background:

• Sedimentation in open channels causes significant operational costs and requires accurate flow velocity calculations.
• Existing sediment transport models often lack robustness due to limited data ranges.
• Machine learning offers a promising approach to develop more reliable sediment transport models.

Purpose of the Study:

• To develop and evaluate machine learning models for calculating flow velocity in open channels, addressing sedimentation issues.
• To compare the performance of tree-based machine learning algorithms against traditional regression equations.
• To utilize a comprehensive dataset covering a wide range of hydraulic and sediment parameters.

Main Methods:

• Implementation of two tree-based machine learning algorithms: M5 rule tree (M5RT) and M5 regression tree (M5RGT).
• Development of models using six diverse datasets encompassing variations in pipe size, sediment concentration, channel slope, sediment size, and flow depth.
• Comparison of machine learning model performance with established regression equations.

Main Results:

• Machine learning approaches demonstrated superior performance compared to traditional regression models.
• M5RT achieved an RMSE of 1.184, while M5RGT achieved an RMSE of 1.071, indicating high accuracy.
• The tree-based algorithms provided satisfactory and reliable results for sediment transport computation.

Conclusions:

• Tree-based machine learning algorithms, M5RT and M5RGT, are effective tools for calculating flow velocity and managing sediment transport in open channels.
• The study highlights the advantage of using diverse datasets for robust model development.
• These algorithms offer a practical and accurate solution for mitigating sedimentation problems in hydraulic engineering.