ROLE OF INPUT WEIGHTING AND BOUNDING IN THE CHARACTERISTIC GPC METHOD

Journal article

Transcribing the Characteristic decomposition of a transfer function matrix of a discrete-time multivariable system from the frequency domain to the time domain provides a convenient means of breaking down complex multivariable GPC problems into a set of simple scalar Characteristic Generalized Predictive Control (GPC) problems. The difficulty is however, that in so doing one ends up costing the activity of a 'fictitious' set of inputs in place of the inputs into the actual systems. It is the purpose of the present paper to overcome this difficulty and to propose suitable extensions. Another aspect of the paper is to consider the means of bounding inputs from 'below' in order to avoid non-minimum phase behaviour (which is inherent in GPC) and from above in order to avoid 'overfiring' which could come into conflict with physical/economic constraints. The paper proposes some suitable extensions that overcome these problems. The efficacy of the results of the paper are demonstrated by means of two illustrative examples.

Authors: Kouvaritakis, B; Rossiter, J

• Publication status: Published
• Journal: INTERNATIONAL JOURNAL OF CONTROL
• Volume: 51
• Issue: 2
• Pages: 391-419
• Publication date: 1990-02-01
• DOI: 10.1080/00207179008934072
• EISSN: 1366-5820
• ISSN: 0020-7179
• Language: English