Cybernetics And Systems Analysis logo
Editorial Board Announcements Abstracts Authors Archive
Cybernetics And Systems Analysis
International Theoretical Science Journal
-->

UDC 519.2, 519.61, 519.71
V.F. Gubarev1, N.N. Salnikov2, S.V. Melnychuk3


1 Space Research Institute of the National Academy of Sciences
of Ukraine and State Space Agency of Ukraine, Kyiv, Ukraine

v.f.gubarev@gmail.com

2 Space Research Institute of the National Academy of Sciences
of Ukraine and State Space Agency of Ukraine, Kyiv, Ukraine

salnikov.nikolai@gmail.com

3 Space Research Institute of the National Academy of Sciences
of Ukraine and State Space Agency of Ukraine, Kyiv, Ukraine

sergvik@ukr.net

IDENTIFICATION OF REGULARIZED MODELS IN THE LINEAR REGRESSION CLASS

Abstract. Identification of complex discrete systems in the class of linear regression models is considered. The problem of identifying an exact model on noisy initial data is known to be ill-posed. This is especially significant when using high-dimensional models. Within the guaranteed approach to uncertainty used in the article, it is proposed to use the dimension of the model as a regularization parameter. Here we suggested and tested two techniques for estimating optimal dimension and parameters of linear regression model to ensure its consistency in accuracy with the data error. Numerical simulations were carried out and their efficiency was evaluated.

Keywords: identification, linear regression, complex system, regularization, model dimension estimation, SVD, simulation.



FULL TEXT

REFERENCES

  1. Ljung L. System identification: Theory for the user. Prentice-Hall, 1999. 609 р.

  2. Soderstorm T., Stoica P. System identification. Hemel Hempstead, U.K.: Prentice-Hall International, 1989. 612 p.

  3. Van Overschee P., De Moor B. Subspace algorithms for the stochastic identification problem. Automatica. 1993. Vol. 29, N 3. P. 649–660.

  4. Van Overschee P., De Moor B. Subspace identification for linear systems. Boston; London: Kluwer Academic Publishers, 1996. 254 р.

  5. Zhang E., Pintelon R. Identification of multivariable dynamic errors-in-variables system with arbitrary inputs. Automatica. 2017. Vol. 82, N 8. P. 69–78.

  6. Soderstorm T. Errors-in-variables methods in system identification. Cham: Springer, 2018. 485 p.

  7. Hong M., Soderstorm T., Zheng W.X. Accuracy analysis of bias-eliminating least square estimates for errors-in-variables identification. Automatica. 2007. Vol. 43, N 9. P. 1590–1596.

  8. Soderstorm T. Accuracy analysis of the Frisch scheme for identifying errors-in-variables systems. IEEE Transaction on Automatic Control. 2007. Vol. 52, Iss. 6. P. 985–997.

  9. Soderstorm T. Errors-in-variables methods in system identification. Automatica. 2007. Vol. 43, N 6. P. 939–958.

  10. Kleiberg D., Soderstorm T., Yang-Wallentin F. Errors-in-variables system identification using structural equation modeling. Automatica. 2016. Vol. 66, N 4. P. 218–230.

  11. Hyundeok Kang, Guoxiang Gu, Wei Xing Zheng. A graph subspace approach to system identification based on errors-in-variables system models. Automatica. 2019. Vol. 109, N 11. Article 108535.

  12. Norton J.P. Identification of parameter bounds for ARMAX models from records with bonded noise. International Journal of Control. 1987. Vol. 45, N 2. P. 375–390.

  13. Kuntsevich V.M., Lychak M.M., Nikitenko A.S. Solving a system of linear equations with uncertainty in both parts. Cybernetics. 1988. N 4. P. 47–52.

  14. Veres S.M., Norton J.P. Parameter-bounding algorithms for linear errors-in-variables models. Milanese M., Norton J., Piet-Lahanier H., Walter А. (Eds.). Bounding Approaches to System Identification. Boston, MA: Springer, 1996. P. 275–288.

  15. Cerone V. Parameter bounds for ARMAX models from records with bounded errors in variables. International Journal of Control. 1993. Vol. 57, N 1. P. 225–235.

  16. Nazin S.A., Polyak B.T. Parametric estimation by the method of ellipsoids in linear multidimensional systems with an indefinite description of the model. Avtomatika i telemekhanika. 2007. N 6. P. 67–80.

  17. Cerone V., Regruto D. Set-membership error-in-variables identification through convex relaxation techniques. IEEE Transaction on Automatic Control. 2012. Vol. AC-57, N 2. P. 517–522.

  18. Casini M., Garulli A., Vicino A. Feasible parameter set approximation for linear models with bounded uncertain regressors. IEEE Transaction on Automatic Control. 2014. Vol. AC–59, N 11. P. 2910–2920.

  19. Kremenetskiy I.A., Salnikov N.N. Nonstochastic approach to determining the dimension and parameters of linear autoregressive models by the input and output variables measurements. Journal of Automation and Information Sciences. 2010. Vol. 42, N 1. P. 19–31.

  20. Salnikov N.N. Accuracy of estimation of parameters of linear regression on errors in variables. Journal of Automation and Information Sciences. 2010. Vol. 42, N 11. P. 18–30.

  21. Salnikov N.N., Siryk S.V. Parameter estimation algorithm of the linear regression with bounded noise in measurements of all variables. Journal of Automation and Information Sciences. 2013. Vol. 45, N 4. P. 1–15.

  22. Tikhonov A.N., Arsenin V.Ya. Methods for solving ill-posed problems [in Russian]. Moscow: Nauka, 1974. 430 p.

  23. Gubarev V.F., Fatenko V.V. Geometric and variational methods of model order reduction. Comparative analysis. Problemy upravleniya i informatiki. 2018. N 1. P. 38–52.

  24. Gubarev V.F. Model and identification of complex systems [in Ukrainian]. Kyiv: Nauk. dumka, 2014. 248 p.

  25. Golub J., Van Lone Ch. Matrix calculations [Russian translation]. Moscow: Mir, 1999. 548 p.

  26. Soverini U., Soderstorm T. Frequency domain EIV identification combining the Frisch scheme and Yule–Walker equations. Proc. of European Control Conf. (Lin, Austria, 2015). P. 2038–2043.

  27. Rawlings J.B., Mayne D.Q. Model predictive control: Theory and design. Nob Hill Publishing, Madison, WI, 2009.




© 2021 Kibernetika.org. All rights reserved.