Please use this identifier to cite or link to this item:
Title: Characterization of pile stiffness using Artificial Neural Networks
Authors: Quevedo Reina, Román 
Álamo Meneses, Guillermo Manuel 
Padrón Hernández, Luis Alberto 
Aznárez González, Juan José 
Maeso Fortuny, Orlando Fco 
UNESCO Clasification: Materias
Issue Date: 2022
Publisher: International Center for Numerical Methods in Engineering (CIMNE) 
Conference: Congress on Numerical Methods in Engineering (CMN 2022) 
Abstract: In order to determine the flexibility of pile foundations, appropiate models that include the soil-pile interaction mechanism should be considered. These types of models are usually complex and involve a high computational cost, making it difficult to transfer knowledge to other applications. In the literature, simplified expressions (e.g. [1, 2]) have been proposed to evaluate the stiffness of the piles in an efficient way, admitting some uncertainty in the result. The objective of this work is to build a surrogate model based on artificial neural networks (ANN) capable of predicting the stiffness of a pile foundation. A dataset is generated to train the ANN. This synthetic data include the variables that define the foundation and the surrounding soil, and the foundation stiffness, which is evaluated through a previously developed continuous formulation [3]. Comparing the ANN predictions with the results obtained through the numerical tool, its potential to act as surrogate model is observed. The proposed ANN can be used to efficiently estimate the flexibility of pile foundation, without a significant loss of accuracy with respect to rigorous soil-pile interaction models. This work has been performed within financial support from research project PID2020- 120102RB-I00, funded by the Agencial Estatal de Investigaci´on of Spain, MCIN/AEI/ 10.13039/501100011033.
ISBN: 978-84-123222-9-3
Source: Congress on Numerical Methods in Engineering (CMN 2022), p. 280
Appears in Collections:Actas de congresos
Adobe PDF (90,63 kB)
Show full item record

Google ScholarTM




Export metadata

Items in accedaCRIS are protected by copyright, with all rights reserved, unless otherwise indicated.