Scholarlink

<p>The seismic response of tapered piles was investigated using a predictive framework based on radial basis function (RBF) interpolation. A database of 3,600 nonlinear three-dimensional simulations was generated to represent soil–pile interaction under seismic loading. The simulations incorporated nonlinear soil behavior, realistic tapered pile geometry, interface interaction, and dynamic ground motion input. Key parameters, including pile slenderness ratio, taper angle, soil unit weight, and peak ground acceleration, were systematically varied to cover a broad range of design scenarios. Maximum shaft friction, maximum bending moment, and lateral pile-head displacement were extracted and used to develop three predictive models. The models demonstrated high accuracy, with coefficients of determination of 0.990 for shaft friction, 0.997 for bending moment, and 0.991 for displacement, and root-mean-square errors of 0.766&#xa0;MPa, 0.821 MN·m, and 0.156&#xa0;mm, respectively. Gaussian kernels provided the best performance, and 95% prediction intervals showed strong calibration. Feature-importance analysis revealed that soil unit weight governs shaft friction, slenderness ratio controls bending moment, and lateral displacement results from coupled soil–geometry effects. The proposed framework closely reproduces nonlinear simulation results while significantly reducing computational cost, enabling rapid seismic evaluation and preliminary design of tapered pile foundations.</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

A Hybrid Data-Driven Approach for Seismic Response Prediction of Tapered Piles Using RBF Interpolation

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月2日

查看全文
  • Musab Aied Qissab Al-Janabi,
  • Duaa Al-Jeznawi,
  • Hasan Ali Abbas,
  • Sajjad E. Rasheed,
  • Luís Filipe Almeida Bernardo,
  • Hugo Alexandre Silva Pinto

摘要

The seismic response of tapered piles was investigated using a predictive framework based on radial basis function (RBF) interpolation. A database of 3,600 nonlinear three-dimensional simulations was generated to represent soil–pile interaction under seismic loading. The simulations incorporated nonlinear soil behavior, realistic tapered pile geometry, interface interaction, and dynamic ground motion input. Key parameters, including pile slenderness ratio, taper angle, soil unit weight, and peak ground acceleration, were systematically varied to cover a broad range of design scenarios. Maximum shaft friction, maximum bending moment, and lateral pile-head displacement were extracted and used to develop three predictive models. The models demonstrated high accuracy, with coefficients of determination of 0.990 for shaft friction, 0.997 for bending moment, and 0.991 for displacement, and root-mean-square errors of 0.766 MPa, 0.821 MN·m, and 0.156 mm, respectively. Gaussian kernels provided the best performance, and 95% prediction intervals showed strong calibration. Feature-importance analysis revealed that soil unit weight governs shaft friction, slenderness ratio controls bending moment, and lateral displacement results from coupled soil–geometry effects. The proposed framework closely reproduces nonlinear simulation results while significantly reducing computational cost, enabling rapid seismic evaluation and preliminary design of tapered pile foundations.