Scholarlink

<p>Spheroidal weathering, which rounds rock block edges, significantly influences the stability of rock slopes prone to block toppling failure. This study introduces an analytical model that incorporates the effect of edge rounding through the ratio of radius of curvature to block thickness (r/t<sub>b</sub>). Modified limit equilibrium analysis shows that while edge rounding minimally affects sliding, it substantially increases toppling susceptibility, reducing the factor of safety (FS) from 0.97 for sharp-edged blocks to 0.68 at an r/tb ratio of 0.3. To enable fast and accurate stability predictions, a shallow Artificial Neural Network (ANN) model with a 3-64-32-1 feed-forward architecture was developed using three input parameters: r/tb, failure plane angle (α), and friction angle (ϕ). The ANN, trained using the Adam optimizer, demonstrated superior convergence compared to SGD and RMSprop, achieving a coefficient of determination (R²) of 0.953, a root mean square error (RMSE) of 0.067, and a mean absolute error (MAE) of 0.054 on the test dataset. Validation through five-fold cross-validation yielded a mean R² of 0.948 with a standard deviation of ± 0.006, and further assessment using external datasets confirmed its generalizability, with R² values ranging from 0.82 to 0.913. Comparative analysis with Random Forest, SVR, kNN, and MLR confirmed the ANN model’s superior performance. Sensitivity analysis revealed that the friction angle (ϕ) had the highest influence on FS. Monte Carlo-based reliability analysis yielded a reliability index of 2.78 and a corresponding failure probability of 0.00272. A user-friendly interface has been developed for practical implementation.</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

Analytical and predictive assessment of rounding effects on rock slope stability under block toppling using adaptive moment estimation optimization

  • 期刊论文

  • 发表日期：

    2025年7月16日

查看全文
  • Neeraj Chaudhary,
  • Kumar Shubham,
  • Subhadeep Metya,
  • Keshav Kumar Sharma

摘要

Spheroidal weathering, which rounds rock block edges, significantly influences the stability of rock slopes prone to block toppling failure. This study introduces an analytical model that incorporates the effect of edge rounding through the ratio of radius of curvature to block thickness (r/tb). Modified limit equilibrium analysis shows that while edge rounding minimally affects sliding, it substantially increases toppling susceptibility, reducing the factor of safety (FS) from 0.97 for sharp-edged blocks to 0.68 at an r/tb ratio of 0.3. To enable fast and accurate stability predictions, a shallow Artificial Neural Network (ANN) model with a 3-64-32-1 feed-forward architecture was developed using three input parameters: r/tb, failure plane angle (α), and friction angle (ϕ). The ANN, trained using the Adam optimizer, demonstrated superior convergence compared to SGD and RMSprop, achieving a coefficient of determination (R²) of 0.953, a root mean square error (RMSE) of 0.067, and a mean absolute error (MAE) of 0.054 on the test dataset. Validation through five-fold cross-validation yielded a mean R² of 0.948 with a standard deviation of ± 0.006, and further assessment using external datasets confirmed its generalizability, with R² values ranging from 0.82 to 0.913. Comparative analysis with Random Forest, SVR, kNN, and MLR confirmed the ANN model’s superior performance. Sensitivity analysis revealed that the friction angle (ϕ) had the highest influence on FS. Monte Carlo-based reliability analysis yielded a reliability index of 2.78 and a corresponding failure probability of 0.00272. A user-friendly interface has been developed for practical implementation.