A reduced order controller design method for multi-input, multi-output plants based on the Youla parameterization of all stabilizing controllers

Efforts in designing controllers of order less than n-m-l+1 for systems modeled by n state, m input, l output minimal state space realizations are presented. It is shown that manipulations within the Youla parameterization of all stabilizing controllers yield controllers with order equal to the system observability index υ₀ that can generate any desired linear combination of the system state variables. It is possible to place the closed loop poles for a minimal realization with such a linear combination, so the method may be used for controller design. Because υ₀, is often less than the plant order for large multiinput-multioutput (MIMO) plant realizations, the method may prove to be a viable reduced order controller design tool. The technique is used on a tenth order model of an existing NASA test facility to generate a fifth order stabilizing controller