Self-tuning control feedforward for multivariable systems with arbitrary interactor matrix

A general multivariable self tuning controller based on the generalized minimum variance strategy is proposed to control a plant with arbitrary interactor matrix and subject to measurable disturbances plus stochastic disturbances. The way of choosing the weighting polynomial matrices in the performance index is investigated to assign the closed-loop poles at the prescribed positions and achieve the best rejection of measurable disturbances acting on the controlled plant. Moreover, some techniques of updating on-line the weighting polynomial matrices are presented and the decoupling self-tuning control algorithms are obtained for a class of multivariable systems. Simulated examples are described which illustrate the effectiveness of the proposed control algorithms.