Novel DNNs for Stiff ODEs with Applications to Chemically Reacting Flows

04/01/2021

∙

Chemically reacting flows are common in engineering, such as hypersonic flow, combustion, explosions, manufacturing processes and environmental assessments. For combustion, the number of reactions can be significant (over 100) and due to the very large CPU requirements of chemical reactions (over 99 number of flow and combustion problems are presently beyond the capabilities of even the largest supercomputers. Motivated by this, novel Deep Neural Networks (DNNs) are introduced to approximate stiff ODEs. Two approaches are compared, i.e., either learn the solution or the derivative of the solution to these ODEs. These DNNs are applied to multiple species and reactions common in chemically reacting flows. Experimental results show that it is helpful to account for the physical properties of species while designing DNNs. The proposed approach is shown to generalize well.

READ FULL TEXT

Novel DNNs for Stiff ODEs with Applications to Chemically Reacting Flows

NINNs: Nudging Induced Neural Networks

Is it conceivable that neurogenesis, neural Darwinism, and species evolution could all serve as inspiration for the creation of evolutionary deep neural networks?

Deep learning fluid flow reconstruction around arbitrary two-dimensional objects from sparse sensors using conformal mappings

Impatient DNNs - Deep Neural Networks with Dynamic Time Budgets

Adversarial Zoom Lens: A Novel Physical-World Attack to DNNs

Libraries of hidden layer activity patterns can lead to better understanding of operating principles of deep neural networks

Electrochemical transport modelling and open-source simulation of pore-scale solid-liquid systems

Novel DNNs for Stiff ODEs with Applications to Chemically Reacting Flows

Related Research

NINNs: Nudging Induced Neural Networks

Is it conceivable that neurogenesis, neural Darwinism, and species evolution could all serve as inspiration for the creation of evolutionary deep neural networks?

Deep learning fluid flow reconstruction around arbitrary two-dimensional objects from sparse sensors using conformal mappings

Impatient DNNs - Deep Neural Networks with Dynamic Time Budgets

Adversarial Zoom Lens: A Novel Physical-World Attack to DNNs

Libraries of hidden layer activity patterns can lead to better understanding of operating principles of deep neural networks

Electrochemical transport modelling and open-source simulation of pore-scale solid-liquid systems