Loop transfer recovery via a proportional-integral observer

Abstract

A systematic loop transfer recovery (LTR) design procedure based on a proportional-integral (PI) observer is described. In this method the robustness of the optimal regulator is recovered perfectly in the sense that the loop transfer characteristic of the optimal regulator is recovered for all values of a complex variable. An important advantage over previous LTR techniques is that the solution for observer design parameters is found from an underdetermined system of linear equations having two degrees of freedom (in the single-output case). This freedom can be used to shape the response, thereby eliminating the need to improve response properties via model-following techniques. A design example is given to illustrate the method