Scholarlink

<p>Numerical models and machine learning methods are implemented and compared to simulate and predict erosional dam-break flows and bed morphodynamics. The nonlinear shallow water equations, including sediment transport and bedload terms, are solved using a well-balanced finite volume method. Empirical erosion formulas are applied, and the obtained data train and test machine learning models. A comparative study using both computational hydraulics and various machine learning models is presented to simulate and predict erosion flows and bed deformations. The methodology is tested for a dam-break problem over an erodible bed, and results are validated against experimental measurements. The performance of various models, including Bayesian Neural Networks (BNN), K-Nearest Neighbors (KNN), M5 Trees, Multivariate Adaptive Regression Splines (MARS), Multiple Linear Regression (MLR), and Support Vector Machines (SVM), are evaluated in predicting changes in bed and free-surface profiles. In the present study, quantitative evaluations using the coefficient of determination <InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="40808_2024_2215_Article_IEq1.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="21" /> </InlineMediaObject> <EquationSource Format="TEX">\(R^2\)</EquationSource> <EquationSource Format="MATHML"><math> <msup> <mi>R</mi> <mn>2</mn> </msup> </math></EquationSource> </InlineEquation>, Nash–Sutcliffe Efficiency (NSE), and normalized Root Mean Square Error (nRMSE) revealed that SVM (with <InlineEquation ID="IEq2"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="40808_2024_2215_Article_IEq2.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="71" /> </InlineMediaObject> <EquationSource Format="TEX">\(R^2 = 0.99\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msup> <mi>R</mi> <mn>2</mn> </msup> <mo>=</mo> <mn>0.99</mn> </mrow> </math></EquationSource> </InlineEquation>, NSE = 0.99, nRMSE = 0.0245) and BNN (with <InlineEquation ID="IEq3"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="40808_2024_2215_Article_IEq3.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="71" /> </InlineMediaObject> <EquationSource Format="TEX">\(R^2 = 0.98\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msup> <mi>R</mi> <mn>2</mn> </msup> <mo>=</mo> <mn>0.98</mn> </mrow> </math></EquationSource> </InlineEquation>, NSE = 0.98, nRMSE = 0.035) significantly outperformed other models, with SVM slightly better during validation and testing processes. This methodology optimizes the existing empirical models with machine learning and therefore, improving the prediction reliability for erosional dam-break flows. These findings are very important for hydraulics engineering by providing improved tools for accurate modelling and efficient simulation of sediment transport problems and thus have the potential to support practical applications in the field.</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

Predicting morphodynamics in dam-break flows using combined machine learning and numerical modelling

  • 期刊论文

  • 发表日期：

    2025年1月11日

查看全文
  • Alia Al-Ghosoun,
  • Veysel Gumus,
  • Mohammed Seaid,
  • Oguz Simsek

摘要

Numerical models and machine learning methods are implemented and compared to simulate and predict erosional dam-break flows and bed morphodynamics. The nonlinear shallow water equations, including sediment transport and bedload terms, are solved using a well-balanced finite volume method. Empirical erosion formulas are applied, and the obtained data train and test machine learning models. A comparative study using both computational hydraulics and various machine learning models is presented to simulate and predict erosion flows and bed deformations. The methodology is tested for a dam-break problem over an erodible bed, and results are validated against experimental measurements. The performance of various models, including Bayesian Neural Networks (BNN), K-Nearest Neighbors (KNN), M5 Trees, Multivariate Adaptive Regression Splines (MARS), Multiple Linear Regression (MLR), and Support Vector Machines (SVM), are evaluated in predicting changes in bed and free-surface profiles. In the present study, quantitative evaluations using the coefficient of determination \(R^2\) R 2 , Nash–Sutcliffe Efficiency (NSE), and normalized Root Mean Square Error (nRMSE) revealed that SVM (with \(R^2 = 0.99\) R 2 = 0.99 , NSE = 0.99, nRMSE = 0.0245) and BNN (with \(R^2 = 0.98\) R 2 = 0.98 , NSE = 0.98, nRMSE = 0.035) significantly outperformed other models, with SVM slightly better during validation and testing processes. This methodology optimizes the existing empirical models with machine learning and therefore, improving the prediction reliability for erosional dam-break flows. These findings are very important for hydraulics engineering by providing improved tools for accurate modelling and efficient simulation of sediment transport problems and thus have the potential to support practical applications in the field.