Order Reduction via Multiple Singular Parameters

Model reduction methods based upon singular perturbation theory is now well developed in analysis and design of large scale control systems. However, applications of singular-perturbational results are dictated by the fact that mathematical models must be expressed in the singularly perturbed form. The problem of identifying singular parameters from a dynamic model of a large scale physical system is still relatively unexplored. Sanuti, Syrcos and Wason recently made a significant contribution in developing a systematic procedure which converts the state equations of a large scale linear time-invariant system to singularly perturbed form. [1][2] It is the purpose of this paper to address the problem of identifying singular parameters which will facilitate controller and observer designs with the aid of reduced-order models. In this paper singular values of an internally-balanced generalized state model are shown to be used as singular parameters in multiparameter singular-perturbational models.