Novel observer/controller identification method-based minimal realisations in block observable/controllable canonical forms and compensation improvement. Issue 7 (19th May 2017)

• Record Type: Journal Article
• Authors: Wu, Chen-Yin; Tsai, Jason Sheng-Hong; Guo, Shu-Mei; Su, Te-Jen; Shieh, Leang-San; Yan, Jun-Juh
• Is Part Of: International journal of systems science. Volume 48:Issue 7(2017)
• Page Start: 1522
• Page End: 1536
• Publication Date: 2017-05-19
• Subjects: OKID -- OCID -- minimal realisation -- canonical form -- optimal tracker -- system identification; System analysis -- Periodicals; 003.3
• DOI: 10.1080/00207721.2016.1269221
• Languages: English
• ISSNs: 0020-7721

Abstract

ABSTRACT: This paper proposes a novel observer/controller identification method for identifying the minimally realised equivalent (reduced-order) mathematical models in the block observer/controller-canonical forms of the unknown (i) open-loop system, (ii) existing feedback/feedforward controllers and/or (iii) observer, based on available measurements of the operating closed-loop system. By skipping the singular value decomposition procedure and without involving the model conversion of the identified model from the general coordinate into the block observer/controller-canonical forms during the identification process, the proposed method is able to directly realise the identified parameters in the minimally realised block observer/controller-canonical forms. This simplifies the system identification process. The new procedures enable us to enhance the computational aspects of designing self-tuning controllers for online adaptive control of (a class of) multivariable systems and to improve the tracking performance considerably. As a result, the newly proposed compensation improvement approach is able to compensate the undesirable operating controller.