A new decoupling design of self-tuning multivariable generalized predictive control

Abstract

A novel robust decoupling method with multivariable generalized predictive control (MGPC) for a class of nonlinear systems is presented in an adaptive version. The cross-coupling action and the non-linear actors of the system are identified on-line by a neural network. A feedforward compensation based on generalized predictive control, is proposed for decoupling control. A modified recursive least-squares (RLS) algorithm can be used to estimate the linear parameters for time-varying systems. Simulations are carried out and the results show the effectiveness of the proposed algorithm. Copyright © 1999 John Wiley & Sons, Ltd.