Scholarlink

<p>The objective of this study is to accurately evaluate pile bearing capacity without the use of sophisticated finite element software or wave recording by utilizing artificial neural networks and public pile driving analysis (PDA) test data. The data from 46 concrete piles in Bandar Imam and 65 steel piles in Asaluyeh will be examined using reliable tests. This article used PDA test data and a system made up of two kinds of artificial neural networks to achieve acceptable results for the pile capacity after introducing previously established relationships. In this system, the pile capacity for a chosen number of piles was finally determined by successfully combining a number of neural networks. The suggested system extends earlier neural networks for a comparable function and uses a self-organizing neural network for a set of data in training, testing, and valid datasets. Model predictability increased by 80% when parameters like hammer drop height were added. Cross-sectional area and pile length were additional improvements, and a normalized stiffness parameter provided marginal gains. However, better results were obtained when length and area were used independently. In the end, predictions were greatly improved by introducing a Bearing Capacity Index (BI) from classical equations. The suggested model is built on this index and PDA data, which provides the best prediction accuracy and efficiency for figuring out pile load-bearing capacity.</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

Analyzing analytical and software methods for deep foundation analysis and presenting a new solution for determining pile capacity using the PDA test

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年6月20日

查看全文
  • Ahmad Honarjoo,
  • Vahed Ghiasi

摘要

The objective of this study is to accurately evaluate pile bearing capacity without the use of sophisticated finite element software or wave recording by utilizing artificial neural networks and public pile driving analysis (PDA) test data. The data from 46 concrete piles in Bandar Imam and 65 steel piles in Asaluyeh will be examined using reliable tests. This article used PDA test data and a system made up of two kinds of artificial neural networks to achieve acceptable results for the pile capacity after introducing previously established relationships. In this system, the pile capacity for a chosen number of piles was finally determined by successfully combining a number of neural networks. The suggested system extends earlier neural networks for a comparable function and uses a self-organizing neural network for a set of data in training, testing, and valid datasets. Model predictability increased by 80% when parameters like hammer drop height were added. Cross-sectional area and pile length were additional improvements, and a normalized stiffness parameter provided marginal gains. However, better results were obtained when length and area were used independently. In the end, predictions were greatly improved by introducing a Bearing Capacity Index (BI) from classical equations. The suggested model is built on this index and PDA data, which provides the best prediction accuracy and efficiency for figuring out pile load-bearing capacity.