PINN Phase-Field Fracture IGA Pipeline
About This Architecture
Physics-informed neural networks (PINNs) combined with isogeometric analysis (IGA) enable efficient phase-field fracture simulation through a four-phase pipeline. NURBS inputs feed RFF feature extraction and neural mapping via MLP, constructing trial fields evaluated through IGA/NURBS discretization and material field computation. Energy minimization with adaptive load-path control (using Adam, LBFGS, and RPROP optimizers) solves the variational problem with VUKIMS multipatch coupling. This architecture reduces computational cost while maintaining accuracy for complex fracture propagation problems. Fork and customize this diagram on Diagrams.so to adapt the pipeline for your material models, boundary conditions, or optimization strategies.
People also ask
How do physics-informed neural networks accelerate phase-field fracture simulations using isogeometric analysis?
This PINN-IGA pipeline combines NURBS inputs with RFF feature extraction and MLP neural mapping to construct trial fields, then evaluates them through IGA discretization and material field computation. Energy minimization with sequential optimizer switching (Adam→LBFGS→RPROP) and load-path adaptivity via VUKIMS coupling solves the variational fracture problem efficiently.
- Domain:
- Ml Pipeline
- Audience:
- Computational mechanics researchers and engineers implementing physics-informed neural networks for fracture mechanics
Generated by Diagrams.so — AI architecture diagram generator with native Draw.io output. Fork this diagram, remix it, or download as .drawio, PNG, or SVG.
