Enriched physics-informed neural networks for 2D in-plane crack analysis: Theory and MATLAB code

In this paper, a method based on the physics-informed neural networks (PINNs) is presented to analyze 2D in-plane crack problems in linear elastic fracture mechanics. Instead of using discretizing mesh, PINNs-based meshless and can be trained batches discretely sampled collocation points. order capture asymptotic behavior near-tip displacement stress fields, standard PINNs formulation enriched here by including crack-tip functions such that local region modeled accurately without nodal refinement. The trainable parameters are learned satisfy governing partial differential equations cracked solid corresponding boundary conditions. It found incorporation enrichment substantially simpler more feasible than conventional finite element (FEM) or (BEM). present algorithm tested verified several representative benchmark examples for different loading crack-mode types. Numerical results show provide highly accurate intensity factors (SIFs) with only few degrees freedom. For interested readers, self-contained MATLAB code data-sets supplementing paper also provided Supplementary material.