Scholarlink

<p>Conventional solution techniques like Analytical, Experimental and Numerical methods are often time-consuming techniques and based on hypothetical assumptions. To overcome these limitations, this study employs a data driven approach incorporating Machine Learning. The present study investigates the competence of Artificial Neural Network with Multi-Layer Backpropagation algorithm for the assessment of Large Piled Raft Foundation (LPRFs). The dataset for this study is generated through a finite element program. The input parameters considered includes thickness of raft, pile length-to-raft width ratio (<i>L</i><sub><i>p</i></sub><i>/B</i><sub><i>r</i></sub>), pile diameter, pile spacing, and number of piles. The behaviour of LPRFs is assessed by parameters like load carrying capacity, average settlement, differential settlement, maximum bending moment in raft, minimum bending moment in raft, and piled-raft coefficient. Six different ANN-MLBP models have been developed on cloud computing services to check the effect of change in hyperparameters for the prediction of various output parameters of LPRFs. Initially, a sensitivity analysis was carried out using the dataset generated from numerical analyses, which identified the most influential parameters for each algorithm thus refining the dataset and reducing the complexity of the model. The model efficacy was evaluated using the key performance indicators (KPIs). The results confirm that the ANN-MLBP algorithm achieves a satisfactory KPIs for most of the output parameters, but the algorithm still lacks to achieve a higher accuracy. Although, the adoption of some other algorithm may help achieving a higher value of KPIs.</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

Competence Assessment of Multi-layer Backpropagation Based Artificial Neural Network (ANN-MLBP) for the Behavioural Prediction of Large Piled Raft Foundations

  • 期刊论文

  • 发表日期：

    2026年7月2日

查看全文
  • Bony Shashikumar Sharma,
  • Shashank Singh,
  • Shruti J. Shukla,
  • Chandresh H. Solanki

摘要

Conventional solution techniques like Analytical, Experimental and Numerical methods are often time-consuming techniques and based on hypothetical assumptions. To overcome these limitations, this study employs a data driven approach incorporating Machine Learning. The present study investigates the competence of Artificial Neural Network with Multi-Layer Backpropagation algorithm for the assessment of Large Piled Raft Foundation (LPRFs). The dataset for this study is generated through a finite element program. The input parameters considered includes thickness of raft, pile length-to-raft width ratio (Lp/Br), pile diameter, pile spacing, and number of piles. The behaviour of LPRFs is assessed by parameters like load carrying capacity, average settlement, differential settlement, maximum bending moment in raft, minimum bending moment in raft, and piled-raft coefficient. Six different ANN-MLBP models have been developed on cloud computing services to check the effect of change in hyperparameters for the prediction of various output parameters of LPRFs. Initially, a sensitivity analysis was carried out using the dataset generated from numerical analyses, which identified the most influential parameters for each algorithm thus refining the dataset and reducing the complexity of the model. The model efficacy was evaluated using the key performance indicators (KPIs). The results confirm that the ANN-MLBP algorithm achieves a satisfactory KPIs for most of the output parameters, but the algorithm still lacks to achieve a higher accuracy. Although, the adoption of some other algorithm may help achieving a higher value of KPIs.