Identification of second-order Kautz models by two-step pole location optimisation

Laguerre models are a simple solution for orthonormal basis function modelling, but such simplicity is accompanied by limitations to model processes whose poles are sparse and when the system has zeros. In this article conditions for position and ratio optimisation of two real poles are analytically derived and used to develop methods for the estimation of real two-parameter Kautz models. Both the models obtained by means of the proposed methods approximated a process with sparse poles and zero with greater accuracy than the second-order Laguerre model.