Data-driven control and interconnection

Abstract

In this thesis we study the design of controllers from data for a class of linear-time invariant dynamical systems. We formulate and study conditions under which data from a system can be used to compute a controller which meets a given performance specification. Under such conditions we develop algorithms for finding a controller. Our methods do not focus only on finding control inputs trajectories but one can also find a representation of the controller under suitable conditions. We also give necessary and sufficient conditions for determining whether a system is dissipative using data.

Finally, we study the notion of difference flatness for linear systems. We define a flat system, using trajectories, as a system whose variables admits a partition such that a subset of the variables are unconstrained, and the other remaining variables are observable from the unconstrained variables. Using this definition we give a characterization of representations of flat systems. We also study the control of flat systems using the notion of control as interconnection.

More information

Identifiers

Catalogue record

Export record