A Design Method for a Multivariable Generalized Predictive Control with a Terminal Matching Condition in a State-space Approach

Abstract

In this paper, the generalized predictive control (GPC) with a terminal matching condition for single-input single-output (SISO) plants proposed by Kwok and Shah [6] is extended so that it can be applied to multi-input multi-output (MIMO) plants. For the extension to MIMO case, a state-space approach is used in the proposed method while a polynomial approach is used in the conventional method. This is because a state-space approach for the design of the proposed GPC makes the extension to MIMO case possible. In addition, a steady-state predictor originally derived in this paper using a state-space approach is shown to be completely equivalent to that in the conventional method when SISO plant is designed. Finally, a numerical example is demonstrated to show that a terminal matching condition in the MIMO case has the same role for improving performances of the controlled system as in the SISO case using conventional method.