Numerical Methods in Control and Optimization of Dynamical Systems

Abstract

The underlying topic of this lecture is the development, analysis, and implementation of numerical algorithms for robust control and stabilization of (partial) differential equations. A particular focus lies on large-scale descriptor systems that play a role in the control of flows. Within the vast research field of numerical methods for control systems, we pursue the so-called H-infinity controller design that is designed to work even if the model is faulty.

Date
Jan 16, 2023 14:00
Location
TU Berlin

After an introduction into the basic notions and principles of dynamical systems and control, I will address the challenges that come with high-dimensional (or even infinite-dimensional) nonlinear systems and show some recent theoretical and numerical approaches to their solution. The findings and concepts are illustrated by applications to the Navier-Stokes model equations and to a real-world triple pendulum.

If time suffices, I will address data-driven approaches that use data and possibly techniques from Machine Learning to enhance the models in terms of accuracy or faster evaluation times.