Researchers have developed novel physics-informed neural network frameworks for discovering constitutive models in mechanics. One approach focuses on identifying anisotropic yield functions in plasticity by representing them as convex neural networks, trained using force equilibrium losses and validated against benchmark studies. Another framework addresses fully coupled thermomechanics by learning internal energy and dissipation potentials using input convex neural networks, ensuring thermodynamic admissibility and demonstrating accuracy on synthetic and experimental data. AI
IMPACT These frameworks could accelerate the development of more accurate and efficient material models in engineering simulations.
RANK_REASON Two arXiv papers detailing novel physics-informed neural network frameworks for discovering constitutive models in mechanics.
- alphaXiv
- arXiv
- CatalyzeX Code Finder for Papers
- DagsHub
- Gotit.pub
- Hagen Holthusen
- Helmholtz free energy
- Hugging Face
- Input convex neural networks
- internal energy
- ScienceCast
- Convex Neural Representations
- plasticity
- Yield Functions
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
