Dynamic interactors in multivariable process control—I. The general time delay case

Abstract

The interactor matrix is shown to be a useful tool for multivariable controller design. In this paper (part I) the case where noninvertibility of the plant is solely due to time delays is treated. (In part II the above restriction will be relaxed.) First, the interactor is used to classify multiinput-multioutput processes according to their delay structure. Subsequently, all admissible controllers are parametrized via the system's interactor matrix and it is shown that a scaled inverse interactor is the optimal closed loop transfer matrix for various performance measures. The best possible decoupling controller is also derived using the interactor. Finally, in a heat exchanger network application, it is shown that the present work can also be used for selecting manipulated variables.