Receding horizon predictive control based upon the Youla-Kučera parametrisation (Q1383477)

A new type of multivariable predictive controller is presented. A quadratic long-range objective function is minimized subject to constraints on input/output signals in the space of all stabilizing controllers given by the Youla-Kučera parametrization. The resulting predictive controller is piecewise linear, has time-varying coefficients in both the adaptive as well as in the nonadaptive case. Its time-varying part is piecewise constant and has the structure of a finite impulse filter. The stability of the time-varying closed-loop system and choices of free parameters in order to move terminal constraints for important finite sequence signals are investigated. The results are applied to the control of a chemical reactor with consecutive and parallel reactions.