From data and structure to models and controllers

Henk van Waarde

Research output: ThesisThesis fully internal (DIV)

439 Downloads (Pure)

Abstract

Systems and control theory deals with analyzing dynamical systems and shaping their behavior by means of control. Dynamical systems are widespread, and control theory therefore has numerous applications ranging from the control of aircraft and spacecraft to chemical process control. During the last decades, a series of remarkable new control techniques have been developed. The majority of these techniques rely on mathematical models of the to-be-controlled system.

However, the growing complexity of modern engineering systems complicates mathematical modeling. In this thesis, we therefore propose new methods to analyze and control dynamical systems without relying on a given system model. Models are thereby replaced by two other ingredients, namely measured data and system structure.

In the first part of the thesis, we consider the problem of data-driven control. This problem involves the development of controllers for a dynamical system, purely on the basis of data. We consider both stabilizing controllers, and controllers that minimize a given cost function.

Secondly, we focus on networked systems. A networked system is a collection of interconnected dynamical subsystems. For this type of systems, our aim is to reconstruct the interactions between subsystems on the basis of data.

Finally, we consider the problem of assessing controllability of a dynamical system using its structure. We provide conditions under which this is possible for a general class of structured systems.
Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • University of Groningen
Supervisors/Advisors
  • Camlibel, Kanat, Supervisor
  • Tesi, Pietro, Co-supervisor
Award date20-Nov-2020
Place of Publication[Groningen]
Publisher
DOIs
Publication statusPublished - 2020

Cite this