Advanced discretization techniques for hyperelastic neural networks
2023/09/06
In our latest work, advanced spatial and temporal discretization techniques are tailored to hyperelastic material models based on physics-augmented neural networks.
In collaboration with Dr. Marlon Franke and Prof. Dr. Peter Betsch from KIT, we have developed finite element and time discretization methods that are tailored to the use of hyperelastic material models based on physics-augmented neural networks. In particular, a novel 7-variable, Hu-Washizu-like mixed formulation is introduced to alleviate locking and an energy-momentum scheme is adjusted for energy and momentum preserving dynamical simulations.
Congratulations to our doctoral researcher Dominik Klein on the publication of the article in Computer Methods in Applied Mechanics and Engineering!
