Application of Neural Network Surrogate Models for Flow Control

Author(s)

Deda, Tarcisio C. ; Wolf, William R. ; Dawson, Scott T. M.

• Document type: Journal article
• Source: AIAA AVIATION 2023 FORUM; v. N/A, p. 10-pg., 2023-01-01.

Abstract

We present the application of two new control techniques that leverage differentiability of neural network (NN) dynamical models. By assembling such models in closed loop with neural network controllers (NNC), backpropagation can be conducted to optimize the control performance represented by a loss function. Furthermore, the linearization of NN dynamical models through computation of Jacobians is also presented as a tool for obtaining linear models that can be used for control design. The approaches are applied to a cylinder flow at Reynolds number Re = 150 and Mach number Ma = 0.3, where two different actuation setups are tested: a pair of min.ets acting in opposition at the top and bottom of the cylinder; and cylinder rotation imposed by an angular acceleration control input. Results show that the proposed methodology succeeds in attenuating unsteadiness of the lift coefficient. The proposed techniques are also applied to a modified version of the Kuramoto-Sivashinsky equation in an iterative training approach, proposed to refine the estimation of dynamics around equilibrium points. (AU)

• FAPESP's process: 13/08293-7 - CCES - Center for Computational Engineering and Sciences
• Grantee: Munir Salomao Skaf
• Support Opportunities: Research Grants - Research, Innovation and Dissemination Centers - RIDC
• FAPESP's process: 19/19179-7 - Flow control strategies for unsteady flows involving transition and turbulence
• Grantee: Tarcísio Costa Déda Oliveira
• Support Opportunities: Scholarships in Brazil - Doctorate
• FAPESP's process: 21/06448-0 - High-fidelity numerical simulations applied in unsteady aerodynamics, turbulence and aeroacoustics
• Grantee: William Roberto Wolf
• Support Opportunities: Research Grants - Young Investigators Grants - Phase 2
• FAPESP's process: 22/00469-8 - Deep learning strategies applied to closed-loop control of unsteady flows
• Grantee: Tarcísio Costa Déda Oliveira
• Support Opportunities: Scholarships abroad - Research Internship - Doctorate