Interpretable Nonlinear Models of Unsteady Flow Physics

Abstract

Understanding and modeling complex fluid flows is a central challenge across many scientific,technological, and industrial applications. Improved models of engineering flows have the potentialfor transformative impact in diverse domains, including energy, transportation, security,and medicine. Unsteady fluid dynamics are particularly challenging to model because of highdimensionality,nonlinearity, and multi-scale physics in both space and time. The objective ofthe proposed work is to develop a data-driven framework to model unsteady fluid flows, resulting in interpretable reduced-order models that elucidate underlying physical interactions. To achieve this goal, we will leverage recent advances in machine learning and convex optimization, namely the recent sparse identification of nonlinear dynamics (SINDy) architecture. We will model canonical unsteady aerodynamic systems, using SINDy to uncover global parameterized models of the two-dimensional cylinder flow across a range of Reynolds numbers and a pitching and plunging airfoil. The resulting models will be parsimonious and avoid overfitting, with each term in the model corresponding to a particular physical interaction. We anticipate that the present work will be transformative, resulting in minimal interpretable models of unsteady aerodynamic systems that are accurate, efficient, robust to changing conditions and mode deformation, and are related to the underlying rich nonlinear flow physics.

Document Details

Document Type
DoD Grant Award
Publication Date
Apr 09, 2018
Source ID
FA95501810200

Entities

People

  • Steven Brunton

Organizations

  • Air Force Office of Scientific Research
  • United States Air Force
  • University of Washington

Tags

Fields of Study

  • Physics

Readers

  • Computational Fluid Dynamics (CFD)
  • Distributed Systems and Data Platform Development
  • Neural Network Machine Learning.

Technology Areas

  • AI & ML
  • AI & ML - Machine Learning Algorithms
  • Space