Title: Multivariable controller tuning for suppression of closed-loop interaction using frequency response dataset

Authors: Kazuhiro Yubai; Shogo Shinoda; Daisuke Yashiro; Junji Hirai

Addresses: Department of Electrical and Electronic Engineering, Graduate School of Mie University, 1577 Kurimamachiya, Tsu, Mie, 514-8507, Japan ' Department of Electrical and Electronic Engineering, Graduate School of Mie University, 1577 Kurimamachiya, Tsu, Mie, 514-8507, Japan ' Department of Electrical and Electronic Engineering, Graduate School of Mie University, 1577 Kurimamachiya, Tsu, Mie, 514-8507, Japan ' Department of Electrical and Electronic Engineering, Graduate School of Mie University, 1577 Kurimamachiya, Tsu, Mie, 514-8507, Japan

Abstract: This paper proposes a multivariable controller design method using frequency response data. The proposed method evaluates diagonal dominance in a closed-loop manner by shaping singular values and designs a controller by solving an optimisation problem that involves non-convex quadratic matrix inequalities. This optimisation problem is rewritten as linear matrix inequalities (LMI) by introducing iterative LMI constraints. The highly diagonal dominant controller is designed by repeating update of the optimisation problem and convex optimisation. This optimisation algorithm guarantees monotonically convergence of an evaluation value in terms of iteration. The experimental results show the effectiveness of the proposed method in comparison with the conventional method.

Keywords: concave-convex procedure; closed-loop interaction; mutivariable control systems; frequency response; controller tuning; controller design; optimisation; linear matrix inequalities; LMI.

DOI: 10.1504/IJAMECHS.2016.082627

International Journal of Advanced Mechatronic Systems, 2016 Vol.7 No.2, pp.71 - 80

Received: 31 Mar 2016
Accepted: 27 Sep 2016

Published online: 02 Mar 2017 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article