Scholarlink

<p>This study focuses on the 3D flow field around straight groynes using experimental datasets and kernel-based techniques. In this regard, firstly, the three-dimensional turbulent flow field around straight groynes with different lengths was investigated using laboratory experiments. Then, two kernel-based approaches including the Kernel Extreme Learning Machine (KELM) and Gaussian Process Regression (GPR) were used for modeling the three-dimensional velocity components. Finally, the modeling uncertainty was investigated using the Lower Upper Bound Estimation (LUBE) method. Estimation (LUBE) method was applied to explore modeling uncertainty. Experimental results showed that increasing the length of the groyne enhanced the maximum kinetic energy of the turbulent flow along the shear layer while also making the energy distribution in the depth more uniform. The groyne with largest length had the highest general turbulent shear stress in the downstream of the structure. The results showed that the kernel-based methods successfully simulated the 3D velocity components, and the GPR performed much better than the KELM. Based on the sensitivity analysis, the vertical coordinate of the measuring point (<i>Z</i><sup><i>*</i></sup>) was the least important input variable in flow field simulation around a single straight groyne. It was found that the GPR demonstrated a desirable degree of reliability.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Assessing the flow field around single groynes with rectangular head shapes via experimental data and intelligence methods

  • 期刊论文

  • 发表日期：

    2025年8月1日

查看全文
  • Akbar Safarzadeh,
  • Roghayeh Ghasempour,
  • Seyed Mahdi Saghebian

摘要

This study focuses on the 3D flow field around straight groynes using experimental datasets and kernel-based techniques. In this regard, firstly, the three-dimensional turbulent flow field around straight groynes with different lengths was investigated using laboratory experiments. Then, two kernel-based approaches including the Kernel Extreme Learning Machine (KELM) and Gaussian Process Regression (GPR) were used for modeling the three-dimensional velocity components. Finally, the modeling uncertainty was investigated using the Lower Upper Bound Estimation (LUBE) method. Estimation (LUBE) method was applied to explore modeling uncertainty. Experimental results showed that increasing the length of the groyne enhanced the maximum kinetic energy of the turbulent flow along the shear layer while also making the energy distribution in the depth more uniform. The groyne with largest length had the highest general turbulent shear stress in the downstream of the structure. The results showed that the kernel-based methods successfully simulated the 3D velocity components, and the GPR performed much better than the KELM. Based on the sensitivity analysis, the vertical coordinate of the measuring point (Z*) was the least important input variable in flow field simulation around a single straight groyne. It was found that the GPR demonstrated a desirable degree of reliability.