Scholarlink

<p>Assessing ground conditions with regard to the potential for squeezing and the behavior of geological structures is a crucial aspect of designing underground structures. Tunnel construction in conditions where squeezing occurs can be a challenging and time-consuming process. Recognizing and evaluating the potential for squeezing is essential in determining the appropriate excavation method and support system, particularly in weak rock formations. This study introduces an innovative application of multiple artificial intelligence (AI) methods, comparing their effectiveness in accurately classifying tunnel squeezing based on real-world data. Unlike previous works that focused on empirical or single-model approaches, this research systematically evaluates and optimizes six AI-based classifiers (Decision Trees (DT), K-nearest Neighbors (KNN), Neural Networks (NN), Naive Bayes (NB), Support Vector Machines (SVM), and Ensemble Classification (EC)) using a dataset of 114 tunnel cases. The results reveal that the DT model outperforms other classifiers with an accuracy rate of approximately 0.98. This finding is significant because it presents a reliable AI-driven method for classifying tunnel squeezing, offering engineers and designers a robust tool for mitigating risks in tunnel construction.</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

Optimal AI Model for Accurate Classification of Squeezing in Underground Structures

  • 期刊论文

  • 发表日期：

    2025年1月3日

查看全文
  • Hadi Fattahi,
  • Ehsan Mohtarami

摘要

Assessing ground conditions with regard to the potential for squeezing and the behavior of geological structures is a crucial aspect of designing underground structures. Tunnel construction in conditions where squeezing occurs can be a challenging and time-consuming process. Recognizing and evaluating the potential for squeezing is essential in determining the appropriate excavation method and support system, particularly in weak rock formations. This study introduces an innovative application of multiple artificial intelligence (AI) methods, comparing their effectiveness in accurately classifying tunnel squeezing based on real-world data. Unlike previous works that focused on empirical or single-model approaches, this research systematically evaluates and optimizes six AI-based classifiers (Decision Trees (DT), K-nearest Neighbors (KNN), Neural Networks (NN), Naive Bayes (NB), Support Vector Machines (SVM), and Ensemble Classification (EC)) using a dataset of 114 tunnel cases. The results reveal that the DT model outperforms other classifiers with an accuracy rate of approximately 0.98. This finding is significant because it presents a reliable AI-driven method for classifying tunnel squeezing, offering engineers and designers a robust tool for mitigating risks in tunnel construction.