ID 70

Towards Data-driven Control of Multiphase Flows

 

Abstract:

There are a number of engineering applications where the governing system of interest is of very high dimension or not known entirely. However it is quite possible to access physical experimental data for such systems. Complex multiphase flows belong to this class of systems. To design control strategies for such systems it is useful to develop a data-driven control method. As a model equation for our eventual spray control application, we use state-of-the-art data-driven system identification and model reduction techniques to design optimal control for the linear and nonlinear Ginzburg Landau equation. The focus of this work is to develop reduced-order direct and adjoint operators that approximate the state transition operator of the high dimensional system. We then use these operators for optimal control and evaluate the performance with the controllers. Perspectives on applying the data-driven methods to multiphase flow control will also be given.