Machine learning
Deep Neural Networks are capable to learn the dynamical system of Navier-Stokes equations from databases without explicitly feeding the governing equations. The combination of Machine Learning with the Principal Component Analysis/Proper Orthogonal Decomposition allows the reduction of feed-data by orders of magnitude for efficient reduced-order models.
Technology
- Deep Neural Network
- Principal Component Analysis
- Data-driven methods
Galerkin projections
Data-driven method reinforced by physical equations for development of reduced order models.
Technology
- Principal Component Analysis
- Physics reinforced models
