A Physics-informed neural network-based Surrogate Model for Analyzing Elasticity Problems in Plates with Holes

Abstract

The elasticity problem of plates with holes is a classic problem in structural engineering and has significant implications for various industrial applications. Traditional numerical methods, such as finite element (FEM) analysis, require substantial computational resources and expertise. To solve this problem, we propose an innovative approach, a physics-informed neural network (PINN) -based surrogate model for solving elasticity problems in plates with holes. By training the PINN model on a dataset generated from FEM simulations with plates of different holes, we achieve accurate predictions of the stress and deformation fields, eliminating the need for laborious FEM computations. Our results demonstrate that the PINN-based surrogate model offers a computationally efficient and reliable approach for analyzing plates with holes of various sizes.