Development of deep learning-based joint elements for thin-walled beam structures

Jaemin Jeon,Jaeyong Kim,Jong Jun Lee,Dongil Shin,Yoon Young Kim

doi:10.1016/j.compstruc.2021.106714

Jaemin Jeon, Jaeyong Kim + Show 3 more

Open Access

https://doi.org/10.1016/j.compstruc.2021.106714

Copy DOI

Abstract

This study presents a new modeling technique to estimate the stiffness matrix of a thin-walled beam-joint structure using deep learning. When thin-walled beams meet at joints, significant sectional deformations occur, such as warping and distortion. These deformations should be considered in the one-dimensional beam analysis, but it is difficult to explicitly express the coupling relationships between the beams’ deformations connected at the joint. This study constructed a deep learning-based joint model to predict the stiffness matrix of a higher-order one-dimensional super element that presents the relationships. Our proposition trains the neural network using the eigenvalues and eigenvectors of the joint's reduced stiffness matrix to satisfy the correct number of zero-strain energy modes overcoming the randomly perturbed error of the deep learning. The deep learning-based joint model produced compliance errors mostly within 2% for a given structural system and the maximum error of 4% in the worst case. The newly proposed methodology is expected to be widely applicable to structural problems requiring the stiffness of a reduction model.

Highlights

Even though computing power has improved dramatically for the last few decades, low-dimensional beam-based finite element models still receive much attention in the vehicle [1,2,3] and civil industry fields [4,5] because of their design-friendliness
The accuracy of the DLbased joint model was verified for two-beam joint structures shown in Fig. 5(a) and more general 2D and 3D beam structures shown in Figs. 5(b) and 1(a) were validated
This paper proposed a deep learning approach to predict the stiffness of a thin-walled beam joint using one-dimensional super elements

Summary

Introduction

Even though computing power has improved dramatically for the last few decades, low-dimensional beam-based finite element models still receive much attention in the vehicle [1,2,3] and civil industry fields [4,5] because of their design-friendliness. The classical six degrees-of-freedom (DOFs) beam models are inaccurate for thin-walled beam structures. While higher-order beams employing more degrees of freedom than those used in the classical beam models can predict solution behavior correctly, matching the field quantities at a joint of thin-walled beams or deriving the joint stiffness matrix suitable for the matching is difficult. We propose a new deep learning (DL)-based joint model using a higher-order

Objectives

Methods

Results

Conclusion