Zeit: | 6. Juni 2023 |
---|---|
Download als iCal: |
|
Prof. Ivan Markovsky
International Centre for Numerical Methods in Engineering
Barcelona, Spain
Tuesday 2023-06-06 4:00 p.m.
IST Seminar Room 2.255 - Pfaffenwaldring 9 - Campus Stuttgart-Vaihingen
Abstract
The talk gives a self-contained derivation of data-driven methods developed in the behavioral setting and demonstrates their relevance for applications. The methods reviewed combine ideas from subspace identification and machine learning. A key idea from subspace identification is that under a persistency of excitation condition, the image of a Hankel matrix constructed from the data is equal to the behavior of the system. This result allows construction of trajectories directly from data, which in turn allows solving data-driven simulation, smoothing, and control problems. The construction requires solution of a system of linear equations only. It assumes, however, that the data is obtained from a linear time-invariant system. For noisy data and nonlinear systems, sparsity promoting regularization is an effective heuristic. As a show case of the data-driven representation, a direct data-driven method for frequency response estimation is shown.
Biographical Information
Ivan Markovsky is an ICREA professor at the International Centre for Numerical Methods in Engineering, Barcelona. He received his Ph.D. degree in Electrical Engineering from the Katholieke Universiteit Leuven in February 2005. From 2006 to 2012 he was an Assistant Professor at the School of Electronics and Computer Science of the University of Southampton and from 2012 to 2022 an Associate Professor at the Vrije Universiteit Brussel. He is a recipient of an ERC starting grant "Structured low-rank approximation: Theory, algorithms, and applications" 2010--2015, Householder Prize honorable mention 2008, and research mandate by the Vrije Universiteit Brussel research council 2012--2022. His main research interests are computational methods for system theory, identification, and data-driven control in the behavioral setting.