An enhanced graph convolutional network-based method for predicting the mechanical behavior of lattice structures with random defects

Zhewei Tian

doi:10.1016/j.tws.2026.115496

An enhanced graph convolutional network-based method for predicting the mechanical behavior of lattice structures with random defects

2026-08-07

0

PRE

AI

Z

Zhewei Tian

X

Xiaoyun Li

Y

Yijun Chai

X

Xiongwei Yang

Y

Yueming Li *

DOI：10.1016/j.tws.2026.115496

Original

Original request for help

Save

Abstract

En 中文

• An enhanced GCN-based method predicts mechanical behavior of defective lattices. • Randomly distributed defects are captured via structure-aware graph modeling. • The method generalizes well across unseen random defect configurations. • The proposed method serves as an efficient surrogate for defect-sensitive analysis.

Keywords:

Lattice structures

Random defect

Mechanical behavior prediction

Graph convolutional networks

An enhanced graph convolutional network-based method for predicting the mechanical behavior of lattice structures with random defects

Abstract

Journal

Organization

Cited Papers

Citing Papers