Scholarlink

<p>Flow characteristics over stepped spillway for relative energy, relative energy loss, relative height of jump, relative length of roller and relative length of jump were studied to optimize impact of steps dimensions. Experimental analyses for Froude number between 1 to 7 were conducted to study the variation in flow characteristics. Artificial Neural Network (ANN) models for these characteristics were created using the experimentally obtained dataset for comparison and a deeper comprehension of hydraulic jump. Performance evaluation of different ANN models namely BFGS Quasi-Newton (BFG), Conjugate Gradient with Powell/Beale Restarts (CGB), Fletcher-Powell Conjugate Gradient (CGF), Polak-Ribiére Conjugate Gradient (CGP), Gaussian Discriminant Analysis (GDA), Variable Learning Rate Back propagation (GDX), Levenberg-Marquardt (LMA), One Step Secant (OSS), Resilient Back propagation (RB), Scaled Conjugate Gradient (SCG) were also carried out. It is predicted that recirculation and air entrainment cause energy dissipation to rise with the number of steps. LMA and BFG prove to be the best and the worst model to assess flow characteristics. Their corresponding RMSE values are 0.014 and 0.112 respectively. Epochs and gradient error were found to be the lowest for LMA, at 5.89 x 10<sup>-4</sup> and 22 correspondingly. The current work can guide practitioners and researchers in selecting the optimal output prediction model for the dataset with the necessary accuracy.</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

Investigation and Evaluation of Hydraulic Parameters of Stepped Spillway Based on ANN Models

  • 期刊论文

  • 发表日期：

    2025年8月11日

查看全文
  • Prachi Dharmadhikari,
  • Sumit Gandhi

摘要

Flow characteristics over stepped spillway for relative energy, relative energy loss, relative height of jump, relative length of roller and relative length of jump were studied to optimize impact of steps dimensions. Experimental analyses for Froude number between 1 to 7 were conducted to study the variation in flow characteristics. Artificial Neural Network (ANN) models for these characteristics were created using the experimentally obtained dataset for comparison and a deeper comprehension of hydraulic jump. Performance evaluation of different ANN models namely BFGS Quasi-Newton (BFG), Conjugate Gradient with Powell/Beale Restarts (CGB), Fletcher-Powell Conjugate Gradient (CGF), Polak-Ribiére Conjugate Gradient (CGP), Gaussian Discriminant Analysis (GDA), Variable Learning Rate Back propagation (GDX), Levenberg-Marquardt (LMA), One Step Secant (OSS), Resilient Back propagation (RB), Scaled Conjugate Gradient (SCG) were also carried out. It is predicted that recirculation and air entrainment cause energy dissipation to rise with the number of steps. LMA and BFG prove to be the best and the worst model to assess flow characteristics. Their corresponding RMSE values are 0.014 and 0.112 respectively. Epochs and gradient error were found to be the lowest for LMA, at 5.89 x 10-4 and 22 correspondingly. The current work can guide practitioners and researchers in selecting the optimal output prediction model for the dataset with the necessary accuracy.