Affordable Access

Design of $L2$ stable fixed-order decentralised controllers in a network of sampled-data systems with time-delays

Authors
  • Dileep, Deesh
  • Thomas, Jijju
  • Hetel, Laurentiu
  • Wouw, Nathan van de
  • Richard, Jean-Pierre
  • Michiels, Wim
Publication Date
Feb 01, 2020
Source
Kaleidoscope Open Archive
Keywords
Language
English
License
Unknown
External links

Abstract

A methodology is proposed for the design of sampled-data fixed-order decentralised controllers for Multiple Input Multiple Output (MIMO) Linear Time-Invariant (LTI) time-delay systems. Imperfections in the communications links between continuous-time plants ans controllers arising due to transmission time-delay, aperiodic sampling, and asynchronous sensors ans actuators are considered. We model the errors induced due to the control imperfections using an operator approach leading to a simple $L2$, stability criterion. A frequency domain-based direct optimisation approach towards controller desing is proposed in thir paper. This approach relies on the minimisation of cost functions, for stability and robustness against control imperfections, as a function of the controller or design parameters. First, the proposed method towards controller design is applied to generic MIMO LTI systems with time-delay. Second, when the delay system to be controlled has the structure of a network of coupled quasi-identical subsystems, we use a scalable algorithm to design identical decentralised controllers through network structure exploitation. Quasi-identical subsystems are identical subsystems that have non-identical uncertainties or control imperfections. By exploiting the structure, we improve the computational efficiency and scalability with the number of subsystems. The methodology has been implemented in a publicly available software, which supports system models in terms of delay differential algebraic equations. Finally, the effectiveness of the methodology is illustrated using a numerical example.

Report this publication

Statistics

Seen <100 times