Deep learning-based design model for suction caissons on clay

被引:6
作者
Yin, Xilin [1 ]
Wang, Huan [1 ,3 ]
Pisano, Federico [1 ]
Gavin, Ken [1 ]
Askarinejad, Amin [1 ]
Zhou, Hongpeng [2 ]
机构
[1] Delft Univ Technol, Fac Civil Engn & Geosci, NL-2628 CN Delft, Netherlands
[2] Univ Manchetser, Fac Sci & Engn, Dept Comp Sci, Manchetser M13 9PL, England
[3] Norwegian Geotech Inst, Adv Modelling Sect, N-0484 Oslo, Norway
关键词
Caisson foundation; Load-displacement relationship; Deep learning; Numerical modelling; GENERALIZED FAILURE ENVELOPE; PLASTICITY MODEL; SHALLOW FOUNDATIONS; UNDRAINED CAPACITY; BEARING CAPACITY; HYPOPLASTIC MACROELEMENT; SKIRTED FOUNDATIONS; BUCKET FOUNDATIONS; NEURAL-NETWORKS; BEHAVIOR;
D O I
10.1016/j.oceaneng.2023.115542
中图分类号
U6 [水路运输]; P75 [海洋工程];
学科分类号
0814 ; 081505 ; 0824 ; 082401 ;
摘要
Predicting the non-linear loading response is the key to the design of suction caissons. This paper presents a systematic study to explore the applicability of deep learning techniques in foundation design. Firstly, a series of three-dimensional finite element simulations was performed, covering a wide range of embedment ratios and different loading directions, to provide training data for the deep neural network (DNN) model. Then, hyper-parameter tuning was performed and it is found that the basic Fully-Connected (FC) neural network model is sufficient to capture the non-linear response of suction caissons with excellent accuracy and robustness. Furthermore, the optimized FC neural network model was also successfully applied to a database of suction caissons in sand, demonstrating its broad applicability. By comparing three typical DNNs, i.e., FC, Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM), it was observed that the FC neural network model excels over others in terms of simplicity, efficiency and accuracy. More importantly, by looking into the model's generalization performance, the FC neural network model can also identify the change in foundation failure mechanisms. This study demonstrates the DNN's powerful mapping ability and its potential for future use in offshore foundation design.
引用
收藏
页数:14
相关论文
共 72 条
[1]  
Blundell C, 2015, PR MACH LEARN RES, V37, P1613
[2]   The undrained capacity of skirted strip foundations under combined loading [J].
Bransby, M. F. ;
Yun, G. -J. .
GEOTECHNIQUE, 2009, 59 (02) :115-125
[3]   Combined loading of skirted foundations [J].
Bransby, MF ;
Randolph, MF .
GEOTECHNIQUE, 1998, 48 (05) :637-655
[4]  
Brutzkus A, 2017, Arxiv, DOI arXiv:1710.10174
[5]   Standardized sign conventions and notation for generally loaded foundations [J].
Butterfield, R ;
Houlsby, GT ;
Gottardi, G .
GEOTECHNIQUE, 1997, 47 (05) :1051-1054
[6]  
Byrne B., 2002, Wind Engineering, V26, P145, DOI 10.1260/030952402762056063
[7]  
Byrne B.W., 2000, Ph.D. thesis
[8]   A plasticity model describing caisson behaviour in clay [J].
Cassidy, M. J. ;
Randolph, M. F. ;
Byrne, B. W. .
APPLIED OCEAN RESEARCH, 2006, 28 (05) :345-358
[9]   Modelling of nonlinear dynamic behaviour of a shallow strip foundation with macro-element [J].
Cremer, C ;
Pecker, A ;
Davenne, L .
JOURNAL OF EARTHQUAKE ENGINEERING, 2002, 6 (02) :175-211
[10]   SANICLAY: simple anisotropic clay plasticity model [J].
Dafalias, Yannis F. ;
Manzari, Majid T. ;
Papadimitriou, Achilleas G. .
INTERNATIONAL JOURNAL FOR NUMERICAL AND ANALYTICAL METHODS IN GEOMECHANICS, 2006, 30 (12) :1231-1257