Perfect Steady-State Indirect Control

Abstract

Indirect control is commonly used in industrial applications where the primary controlled variable is not measured. This paper considers the case of “perfect indirect control” where one attempts to control a combination of the available measurements such that there is no effect of disturbances on the primary outputs at steady-state. This is always possible provided the number of measurements is equal to the number of independent variables (inputs plus disturbances). It is further shown how extra measurements may be used to minimize the effect of measurement error. The results in this paper also provide a nice link to previous results on inferential control, perfect disturbance rejection and decoupling (DRD), and self-optimizing control.