Graph convolutional networks-driven multiscale topology optimization base on polygonal coarse-grid elements

doi:10.1016/j.ijmecsci.2025.110500

Graph convolutional networks-driven multiscale topology optimization base on polygonal coarse-grid elements

2025-06-15

0

PRE

AI

DOI：10.1016/j.ijmecsci.2025.110500

Original

Original request for help

Save

Abstract

En 中文

• A data-driven framework enables prediction of numerical base functions for heterogeneous polygonal substructures. • The designed neural network achieves superior prediction accuracy and generalization capability. • The trained surrogate adapts to on-demand objectives without offline reprocessing. • An AI-enhanced design paradigm demonstrates higher efficiency than conventional topology optimization.

Graph convolutional networks-driven multiscale topology optimization base on polygonal coarse-grid elements

Abstract

Journal

Organization