Robust multivariable predictive control: application to an industrial test stand

This paper reports a theoretical extension of multivariable predictive control (MPC). The robustness of an augmented algorithm (/spl alpha/-MPC) for a general M-input N-output system is explored. It is shown that an extra parameter /spl alpha/ in the criterion function can reduce the H/sub /spl infin//-norm of the multivariable sensitivity function, thus improving the disturbance-rejection properties of the closed loop system. This control law is finally applied to a test stand for air conditioning equipment of aircraft with a great improvement of performances regarding the former regulation.