Técnicas de identificação por subespaços, aplicadas a modelos de ordem reduzida com atraso.

Abstract

It is proposed in this dissertation the use of concepts of subspaces in the identification of multivariable models of reduced order with time delays. The developed methodology is based on the estimation of the step response of the system from projections of its input signals and outputs. The problem is divided into two stages, first: estimating models in open loop, then followed by the study of closed loop systems. Finally, the concepts studied are implemented through numerical simulations and practical experiments in real plants.