Identification of multivariable linear systems by fitting frequency responses based on temporal responses to the unit step

Authors

  • Tatiane C Ramalho Exército Brasileiro, Instituto Militar de Engenharia. Rio de Janeiro, RJ, Brasil.
  • Paulo C Pellanda Exército Brasileiro, Instituto Militar de Engenharia. Rio de Janeiro, RJ, Brasil.
  • Roberto Ades Exército Brasileiro, Instituto Militar de Engenharia. Rio de Janeiro, RJ, Brasil.
  • Marcelo A Oliveira Exército Brasileiro, Instituto Militar de Engenharia. Rio de Janeiro, RJ, Brasil.

Abstract

This work presents a method to identify multivariable linear models in the frequency domain based on unit step responses and the optimal tuning of their measured frequency responses. Transfer function matrix polynomials coefficients are analytically computed to minimize a cost function based on a square norm. Better model tunings can be obtained by using this new method in terms of both frequency and steady-state time responses. Three numerical examples are used to evidence some relative advantages of the proposed method.

Downloads

Download data is not yet available.

Published

2021-07-01

How to Cite

Ramalho, T. C., Pellanda, P. C., Ades, R., & Oliveira, M. A. (2021). Identification of multivariable linear systems by fitting frequency responses based on temporal responses to the unit step. Revista Militar De Ciência E Tecnologia, 38(3), 65–81. Retrieved from https://ebrevistas.eb.mil.br/CT/article/view/8656