Scholarlink

This study presents an innovative approach for probabilistic slope stability analysis, integrating Gaussian process regression (GPR) with structured meshing in the finite element method (FEM). The methodology is designed to enhance the accuracy and reliability of slope stability assessments in geotechnical engineering. The process involves three main steps. Firstly, FEM’s structured meshing technique is employed to generate training and test samples for constructing the response surface. Next, GPR is utilized to create a response surface, providing an explicit form that simplifies application to complex slope scenarios. Lastly, Monte Carlo simulation (MCS) utilizes the GPR response surface to assess the system failure probability. The effectiveness of the proposed method was confirmed through validation with two classical slope examples. Comparative analysis of the results demonstrated that the FEM with structured meshing reduces random testing errors and avoids unnecessary spatial variability. The computer simulation results confirm that the proposed method accurately estimates system failure probability using the structured meshing technique and Shear strength reduction method (SSRM) of FEM.

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

System Probabilistic Stability Analysis of Slopes Using Gaussian Process Regression via Structured Meshing of Finite Element Method

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Defang Liu,
  • Mingjie He,
  • Ning Liu,
  • Lingang Gao,
  • Junhao Chen

摘要

This study presents an innovative approach for probabilistic slope stability analysis, integrating Gaussian process regression (GPR) with structured meshing in the finite element method (FEM). The methodology is designed to enhance the accuracy and reliability of slope stability assessments in geotechnical engineering. The process involves three main steps. Firstly, FEM’s structured meshing technique is employed to generate training and test samples for constructing the response surface. Next, GPR is utilized to create a response surface, providing an explicit form that simplifies application to complex slope scenarios. Lastly, Monte Carlo simulation (MCS) utilizes the GPR response surface to assess the system failure probability. The effectiveness of the proposed method was confirmed through validation with two classical slope examples. Comparative analysis of the results demonstrated that the FEM with structured meshing reduces random testing errors and avoids unnecessary spatial variability. The computer simulation results confirm that the proposed method accurately estimates system failure probability using the structured meshing technique and Shear strength reduction method (SSRM) of FEM.