Core Concepts
Proposing a model reduction technique for LPV systems based on fixed-structure controller synthesis tools.
Abstract
The content introduces a novel approach to model reduction for linear parameter varying (LPV) systems by transforming the problem into an equivalent controller synthesis problem. The proposed method allows for the imposition of a desired structure on the reduced model while providing an error bound estimate. The process is applied to a benchmark mechanical system and compared with traditional techniques. The paper outlines the preliminaries, model reduction procedure, application on a mass-spring-damper system, and concludes with numerical results comparing different models in open-loop and closed-loop scenarios.
I. Introduction:
LPV systems are introduced as a generalization of LTI systems.
Challenges in scaling available techniques with high complexity LPV systems are highlighted.
II. Preliminaries:
Description of LPV systems depending on time-varying scheduling parameters.
Introduction to Linear Fractional Transformation (LFT) form for rational dependence on ρ.
III. Model Reduction via Fixed Structure Synthesis:
Proposal to minimize approximation error using fixed-structure controller synthesis techniques.
Conversion of model reduction problem into optimal controller synthesis problem.
IV. Numerical Results:
Application of proposed method on a mass-spring-damper system example.
Comparison of reduced models with and without imposed structure in open-loop and closed-loop scenarios.
Stats
"This work was partially funded by the German Research Fundation (DFG) - project number 413984989."
"Lennart Heeren, Adwait Datar, Antonio Mendez Gonzalez and Herbert Werner are with the Institute of Control Systems, Hamburg University of Technology, 21073 Hamburg, Germany."