Simulation of hyperelastic materials in real-time using Deep Learning
This addresses the problem of real-time simulation for engineering applications, offering a faster alternative to traditional methods like proper orthogonal decomposition (POD), though it is incremental as it builds on existing deep learning techniques.
The paper tackles the computational cost of finite element method (FEM) simulations for hyperelastic materials by introducing U-Mesh, a data-driven method using a U-Net architecture to approximate the non-linear relation between contact forces and displacement fields, achieving very fast simulations with small errors on benchmark examples like a cantilever beam and L-shape.
The finite element method (FEM) is among the most commonly used numerical methods for solving engineering problems. Due to its computational cost, various ideas have been introduced to reduce computation times, such as domain decomposition, parallel computing, adaptive meshing, and model order reduction. In this paper we present U-Mesh: a data-driven method based on a U-Net architecture that approximates the non-linear relation between a contact force and the displacement field computed by a FEM algorithm. We show that deep learning, one of the latest machine learning methods based on artificial neural networks, can enhance computational mechanics through its ability to encode highly non-linear models in a compact form. Our method is applied to two benchmark examples: a cantilever beam and an L-shape subject to moving punctual loads. A comparison between our method and proper orthogonal decomposition (POD) is done through the paper. The results show that U-Mesh can perform very fast simulations on various geometries, mesh resolutions and number of input forces with very small errors.