Model selection through a statistical analysis of the global minimum of a weighted nonlinear least squares cost function

This paper presents a model selection algorithm for the identification of parametric models that are linear in the measurements. It is based on the mean and variance expressions of the global minimum of a weighted nonlinear least squares cost function. The method requires the knowledge of the noise covariance matrix but does not assume that the true model belongs to the model set. Unlike the traditional order estimation methods available in literature, the presented technique allows to detect undermodeling. The theory is illustrated by simulations on signal modeling and system identification problems and by one real measurement example.

[1]  Adriaan van den Bos,et al.  Resolvability of the Parameters of Multiexponentials and Other Sum Models , 1993, IEEE Trans. Signal Process..

[2]  Gerd Vandersteen,et al.  General framework for asymptotic properties of generalized weighted nonlinear least-squares estimators with deterministic and stochastic weighting , 1996, IEEE Trans. Autom. Control..

[3]  Lennart Ljung,et al.  System Identification: Theory for the User , 1987 .

[4]  J. Cadzow,et al.  Signal processing via least squares error modeling , 1990, IEEE ASSP Magazine.

[5]  J. Schoukens,et al.  On the Limits of Order Estimation , 1988 .

[7]  Petre Stoica,et al.  Decentralized Control , 2018, The Control Systems Handbook.

[8]  I. J. Leontaritis,et al.  Model selection and validation methods for non-linear systems , 1987 .

[9]  D. Mitchell Wilkes,et al.  ARMA model order estimation based on the eigenvalues of the covariance matrix , 1993, IEEE Trans. Signal Process..

[10]  Thomas Kailath,et al.  Linear Systems , 1980 .

[11]  B. Picinbono.  Random Signals and Systems , 1993 .

[12]  T. Söderström.  On model structure testing in system identification , 1977 .

[13]  M. Kendall,et al.  Kendall's advanced theory of statistics , 1995 .

[14]  Data based models: an automatic method for model structure determination , 1989 .

[15]  J. Schoukens,et al.  Parametric identification of transfer functions in the frequency domain-a survey , 1994, IEEE Trans. Autom. Control..

[16]  S. A. Billings,et al.  Model validity tests for non-linear signal processing applications , 1991 .

[17]  S. Veres.  Relations between information criteria for model-structure selection Part 1. The role of bayesian model order estimation , 1990 .

[18]  J. Schoukens,et al.  Identification of linear systems in the presence of nonlinear distortions. A frequency domain approach. I. Non-parametric identification , 1995, Proceedings of 1995 34th IEEE Conference on Decision and Control.

[19]  Patrick Guillaume,et al.  Measurement of noise (cross-) power spectra for frequency-domain system identification purposes: large-sample results , 1996 .

[20]  J. Rissanen,et al.  Modeling By Shortest Data Description* , 1978, Autom..

[21]  H. Akaike.  A new look at the statistical model identification , 1974 .

[22]  Rik Pintelon,et al.  Identification of Linear Systems: A Practical Guideline to Accurate Modeling , 1991 .