Scholarlink

<p>The present study shows a comprehensive modelling framework to estimate the infiltration rates in the permeable channels using various soft computing techniques, like Random Forest (RF), Distributed Random Forest (DRF), Deep Neural Network (DNN), Artificial neural network (ANN), Stacked Ensemble, and Gradient Boosting Machine (GBM). The models were developed using input parameters which included base width (b), channel side slope (m), water level (y), sand (%), silt (%), clay (%) and time (T). The efficacy of developed models was evaluated using multiple statistical parameters, scatter plots, Taylor diagrams, and box plot. With a high accuracy GBM model was found to perform better than the other applied models with correlation coefficient (CC) of 0.9868 throughout the testing phase and low error values. Additionally, sensitivity analysis was performed using Cosine Amplitude Method (CAM), and the results showed that time and water level (y) are the most important component influencing infiltration rate in the permeable channels. These results underline that the produced model was reliable enough to take the role of the time-consuming laboratory techniques needed to measure the interaction between soil and water.</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

Approximation of infiltration rates in permeable channels using soft computing techniques

  • 期刊论文

  • 发表日期：

    2026年5月29日

查看全文
  • Parveen Sihag,
  • Yogesh Aggarwal,
  • Ahmed Mohammed Sami Al-Janabi,
  • Divya Singh,
  • Sabih Hashim Alzuhairy

摘要

The present study shows a comprehensive modelling framework to estimate the infiltration rates in the permeable channels using various soft computing techniques, like Random Forest (RF), Distributed Random Forest (DRF), Deep Neural Network (DNN), Artificial neural network (ANN), Stacked Ensemble, and Gradient Boosting Machine (GBM). The models were developed using input parameters which included base width (b), channel side slope (m), water level (y), sand (%), silt (%), clay (%) and time (T). The efficacy of developed models was evaluated using multiple statistical parameters, scatter plots, Taylor diagrams, and box plot. With a high accuracy GBM model was found to perform better than the other applied models with correlation coefficient (CC) of 0.9868 throughout the testing phase and low error values. Additionally, sensitivity analysis was performed using Cosine Amplitude Method (CAM), and the results showed that time and water level (y) are the most important component influencing infiltration rate in the permeable channels. These results underline that the produced model was reliable enough to take the role of the time-consuming laboratory techniques needed to measure the interaction between soil and water.