.png){kind=icon}
返回
Deep surrogate-based GPU-accelerated optimization for wind-resistant shear wall layout
.png){kind=icon}
PRE
AI
DOI：10.1016/j.istruc.2026.111071.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
The growing demand for wind-resistant high-rises necessitates efficient shear wall design, yet traditional manual or simulation-based optimization approaches remain costly. This paper introduces a GPU-accelerated deep surrogate-assisted optimization framework for the automated design of shear wall layouts. Treating the shear wall layout as an image-like structural topology, we employ a ConvNeXt neural network within a multi-task learning framework to simultaneously predict multiple structural performance metrics from a single forward pass. The resulting surrogate demonstrates high fidelity to finite element analysis (R² 0.998). It is then embedded within a genetic algorithm with fully GPU-parallelized fitness evaluations, enabling massive exploration of the design space with orders-of-magnitude computational savings. In a representative 20-story building case study, the proposed framework rapidly identifies code-compliant material-efficient layouts. Our methodology is simulation-driven, operating without relying on any pre-existing manual design database, highlighting its potential for rapid optimization of lateral systems in tall buildings.
期刊
IF:
4.3
论文数:
1.3W
被引数:
2.7W
机构
引用论文
Design-condition-informed shear wall layout design based on graph neural networks基于图神经网络的设计状态信息剪力墙布局设计
PREAI
分享收藏A practical shear wall layout optimization framework for the design of high-rise buildings一种实用的高层建筑剪力墙布置优化设计框架.png)
STRUCTURES
OAAI
STRUCTURESIF4.3
OAAI分享收藏分享收藏
查看更多内容