Scholarlink

<p>This study aims to use the finite element analysis (FEA) method combined with a binary classification machine learning model to predict the success or failure of deep excavation projects. Predicting the stability of excavations is crucial in construction projects, especially for urban structures with significant depth and exposure to various complex geological factors. The research methodology involves applying FEA to simulate soil and excavation wall displacements under different loading scenarios and conditions. Based on the FEA analysis results, observational variables such as depth, the number of shoring layers, and horizontal displacement values were used to train the binary classification machine learning model, with the goal of predicting the success or failure of the excavation. A supervised learning model was deployed to optimize predictions based on real-world data. The analysis results show that the shoring system plays a crucial role in limiting displacement of the excavation wall, particularly at greater depths. When the full shoring system is activated, horizontal displacement is better controlled, whereas the absence of shoring leads to significant increases in barrette wall movement, posing a high risk of failure. The machine learning model achieved high accuracy, with performance metrics such as precision and recall both exceeding 90%, confirming the effectiveness of this approach.</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

Deep Excavation Success Prediction: A Hybrid Approach with FEA and Machine Learning

  • 期刊论文

  • 发表日期：

    2025年2月6日

查看全文
  • Phuong Nguyen Tuan,
  • Tuan Nguyen Anh,
  • Truong Dang Xuan,
  • Hoa Tran Vu Van

摘要

This study aims to use the finite element analysis (FEA) method combined with a binary classification machine learning model to predict the success or failure of deep excavation projects. Predicting the stability of excavations is crucial in construction projects, especially for urban structures with significant depth and exposure to various complex geological factors. The research methodology involves applying FEA to simulate soil and excavation wall displacements under different loading scenarios and conditions. Based on the FEA analysis results, observational variables such as depth, the number of shoring layers, and horizontal displacement values were used to train the binary classification machine learning model, with the goal of predicting the success or failure of the excavation. A supervised learning model was deployed to optimize predictions based on real-world data. The analysis results show that the shoring system plays a crucial role in limiting displacement of the excavation wall, particularly at greater depths. When the full shoring system is activated, horizontal displacement is better controlled, whereas the absence of shoring leads to significant increases in barrette wall movement, posing a high risk of failure. The machine learning model achieved high accuracy, with performance metrics such as precision and recall both exceeding 90%, confirming the effectiveness of this approach.