Generalizable physics-constrained modeling using learning and inference assisted by feature-space engineering
Vishal Srivastava, Karthik Duraisamy
Abstract
We present a data-driven framework that uses very limited data to learn generalizable bypass transition models. Trained on just two flat plate cases, the augmented model is shown to generalize to other flat plate cases and turbomachinery configurations.
Topics & Concepts
InferenceSpace (punctuation)Feature (linguistics)Computer scienceArtificial intelligenceTurbomachineryFeature engineeringFeature vectorMachine learningEngineeringDeep learningMechanical engineeringLinguisticsPhilosophyOperating systemTurbomachinery Performance and OptimizationModel Reduction and Neural NetworksFluid Dynamics and Turbulent Flows