On coprime factorization and minimal realization of transfer function matrices using the pseudo-observability concept

Abstract

A simple computational algorithm is presented for calculating the coprime matrix fraction description and minimal state-space representation of a multivariable linear system specified by a transfer function matrix. The concept of admissibility of pseudo-observability indices is utilized in the algorithm to permit extraction of the most convenient representations, in the sense that the representation elements have the lowest absolute values. A computational example is given to show the feasibility of the suggested method