Finite Element Method-enhanced Neural Network for Forward and Inverse Problems

@article{Meethal2022FiniteEM,
  title={Finite Element Method-enhanced Neural Network for Forward and Inverse Problems},
  author={Rishith Ellath Meethal and Birgit Obst and Mohamed Khalil and Aditya Ghantasala and Anoop Kodakkal and Kai-Uwe Bletzinger and Roland W{\"u}chner},
  journal={ArXiv},
  year={2022},
  volume={abs/2205.08321}
}
We introduce a novel hybrid methodology combining classical finite element methods (FEM) with neural networks to create a well-performing and generalizable surrogate model for forward and inverse problems. The residual from finite element methods and custom loss functions from neural networks are merged to form the algorithm. The Finite Element Method-enhanced Neural Network hybrid model (FEM-NN hybrid) is data-efficient and physics conforming. The proposed methodology can be used for surrogate… 
1 Citations
Neural Network-Based Surrogate Models Applied to Fluid-Structure Interaction Problems
  • D. Arcones, R. Meethal, B. Obst, R. Wüchner
  • Computer Science
    15th World Congress on Computational Mechanics (WCCM-XV) and 8th Asian Pacific Congress on Computational Mechanics (APCOM-VIII)
  • 2022
TLDR
Two cases in which one of the systems is substituted by a neural network-based surrogate model are analyzed and results present improvements in the capacity of the previous method applied to FSI problems.

References

SHOWING 1-10 OF 48 REFERENCES
Learning Mesh-Based Simulation with Graph Networks
TLDR
MeshGraphNets is introduced, a framework for learning mesh-based simulations using graph neural networks that can be trained to pass messages on a mesh graph and to adapt the mesh discretization during forward simulation, and can accurately predict the dynamics of a wide range of physical systems.
Accelerating Eulerian Fluid Simulation With Convolutional Networks
TLDR
This work proposes a data-driven approach that leverages the approximation power of deep-learning with the precision of standard solvers to obtain fast and highly realistic simulations of the Navier-Stokes equations.
Error and uncertainty in modeling and simulation
Physics Informed Deep Learning (Part I): Data-driven Solutions of Nonlinear Partial Differential Equations
TLDR
This two part treatise introduces physics informed neural networks – neural networks that are trained to solve supervised learning tasks while respecting any given law of physics described by general nonlinear partial differential equations and demonstrates how these networks can be used to infer solutions topartial differential equations, and obtain physics-informed surrogate models that are fully differentiable with respect to all input coordinates and free parameters.
NeuroAnimator: fast neural network emulation and control of physics-based models
TLDR
By exploiting the network structure of the NeuroAnimator, a novel approach to creating physically realistic animation that exploits neural networks is proposed, a fast algorithm for learning controllers that enables either physics-based models or their neural network emulators to synthesize motions satisfying prescribed animation goals.
Real‐time simulation of surgery by reduced‐order modeling and X‐FEM techniques
TLDR
A combination with X-FEM methods on a locally superimposed patch is developed for cutting simulation without remeshing to obtain models with very few degrees of freedom that run under real-time constrains even for highly non-linear tissue constitutive equations.
Particle-based methods : fundamentals and applications
Advances in the Particle Finite Element Method (PFEM) for solving coupled problems in engineering.- Advances in Computational Modelling of Multi-Physics in Particle-Fluid Systems.- Large Scale
Theory of the backpropagation neural network
  • R. Hecht-Nielsen
  • Computer Science
    International 1989 Joint Conference on Neural Networks
  • 1989
Adam: A Method for Stochastic Optimization
TLDR
This work introduces Adam, an algorithm for first-order gradient-based optimization of stochastic objective functions, based on adaptive estimates of lower-order moments, and provides a regret bound on the convergence rate that is comparable to the best known results under the online convex optimization framework.
An introduction to computational fluid dynamics - the finite volume method
TLDR
This chapter discusses the development of the Finite Volume Method for Diffusion Problems, a method for solving pressure-Velocity Coupling in Steady Flows problems, and its applications.
...
...