Abstract
The design of the advanced control strategies is usually based on a discrete time model of the
plant. The usual approach has been to use the shift operator q, but this gives numerical difficulties with
the small sampling periods. Many algorithms are better-conditioned numerically using delta operator
implementation than that ones based on shift operator implementation. This study concerns the predictive
control strategy in a state space form because it is attractive during implementation and is particularly
easy to extend to the multivariable case. In view of design parameters, the suggested method is very
simple, and opposing to the conventional discrete time GPC, the algorithm proposed exhibits at high
sampling rates lower computational complexity and better numerical conditioning for unstable and nonminimum
phase systems.
