Scholarlink

<p>This study proposes a hybrid soft computing paradigm of extreme learning machine (ELM) and improved grey wolf optimizer (IGWO) to perform risk analysis (RA) of cantilever retaining walls (CR-walls) under static and seismic conditions. Laboratory tests were performed to determine the soil parameters followed by the utilization of Monte Carlo Simulation (MCS) to automate the process of RA using the ELM-IGWO framework. This study investigates three different stability conditions namely sliding, overturning, and bearing failures for five different wall heights along with three different combinations of co-efficient of variations (COVs) of related soil parameters. The proposed framework’s results were compared to those of six additional hybrid ELMs and four ensemble learning paradigms. The ELM-IGWO framework yields the most accurate estimation for all stability factors with correlation coefficients of 0.9969 and 0.9970 in the training and testing phases, respectively. A graphical user interface (GUI) was developed to effectively perform RA of the CR-wall under static and seismic conditions with different levels of COVs enabling users to estimate failure probabilities for different design parameters, including wall height, analysis type, and seismic coefficient. The developed GUI tool is attached as a supplementary material.</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

Risk Analysis of Cantilever Retaining Walls Under Static and Seismic Conditions Using an Improved Machine Learning Paradigm

  • 期刊论文

  • 发表日期：

    2025年8月7日

查看全文
  • Avinash Kumar,
  • Avijit Burman

摘要

This study proposes a hybrid soft computing paradigm of extreme learning machine (ELM) and improved grey wolf optimizer (IGWO) to perform risk analysis (RA) of cantilever retaining walls (CR-walls) under static and seismic conditions. Laboratory tests were performed to determine the soil parameters followed by the utilization of Monte Carlo Simulation (MCS) to automate the process of RA using the ELM-IGWO framework. This study investigates three different stability conditions namely sliding, overturning, and bearing failures for five different wall heights along with three different combinations of co-efficient of variations (COVs) of related soil parameters. The proposed framework’s results were compared to those of six additional hybrid ELMs and four ensemble learning paradigms. The ELM-IGWO framework yields the most accurate estimation for all stability factors with correlation coefficients of 0.9969 and 0.9970 in the training and testing phases, respectively. A graphical user interface (GUI) was developed to effectively perform RA of the CR-wall under static and seismic conditions with different levels of COVs enabling users to estimate failure probabilities for different design parameters, including wall height, analysis type, and seismic coefficient. The developed GUI tool is attached as a supplementary material.