Controller implementability: a data-driven approach
Optimization and Control
2023-10-25 v1 Systems and Control
Systems and Control
Abstract
We study the controller implementability problem, which seeks to determine if a controller can make the closed-loop behavior of a given plant match that of a desired reference behavior. We establish necessary and sufficient conditions for controller implementability which only rely on raw data. Subsequently, we consider the problem of constructing controllers directly from data. By leveraging the concept of canonical controller, we provide a formula to directly construct controllers that implement plant-compatible reference behaviors using measurements of both reference and plant behaviors.
Keywords
Cite
@article{arxiv.2310.15347,
title = {Controller implementability: a data-driven approach},
author = {Alberto Padoan and Jeremy Coulson and Florian Dörfler},
journal= {arXiv preprint arXiv:2310.15347},
year = {2023}
}