Physics-informed MeshGraphNets (PI-MGNs): Neural finite element solvers for non-stationary and nonlinear simulations on arbitrary meshes

• DOI: 10.1016/j.cma.2024.117102
• Publication Date: 2024-06-15T16:37:59Z
• Authors: Tobias Würth; Niklas Freymuth; Clemens Zimmerling; Gerhard Neumann; Luise Kärger
• Keywords: ddc:004; FOS: Computer and information sciences; Computer Science - Machine Learning; Computer Science - Artificial Intelligence; DATA processing & computer science; Physics-based simulation; 0211 other engineering and technologies; 600; 02 engineering and technology; Graph neural network; Partial differential equations; Surrogate model; 620; 004; Machine Learning (cs.LG); Computational Engineering, Finance, and Science (cs.CE); Artificial Intelligence (cs.AI); Machine learning; Computer Science - Computational Engineering, Finance, and Science; info:eu-repo/classification/ddc/004

Engineering components must meet increasing technological demands in ever shorter development cycles. To face these challenges, a holistic approach is essential that allows for the concurrent development of part design, material system and manufacturing process. Current approaches employ numerical simulations, which quickly becomes computation-intensive, especially for iterative optimization. Data-driven machine learning methods can be used to replace time- and resource-intensive numerical simulations. In particular, MeshGraphNets (MGNs) have shown promising results. They enable fast and accurate predictions on unseen mesh geometries while being fully differentiable for optimization. However, these models rely on large amounts of expensive training data, such as numerical simulations. Physics-informed neural networks (PINNs) offer an opportunity to train neural networks with partial differential equations instead of labeled data, but have not yet been extended to handle time-dependent simulations of arbitrary meshes. This work introduces PI-MGNs, a hybrid approach that combines PINNs and MGNs to quickly and accurately solve non-stationary and nonlinear partial differential equations (PDEs) on arbitrary meshes. The method is exemplified for thermal process simulations of unseen parts with inhomogeneous material distribution. Further results show that the model scales well to large and complex meshes after being trained exclusively on small, generic meshes.