论文信息 - SOME PECULIARITIES OF IDENTIFICATION IN THE PRESENCE OF MODEL ERRORS

SOME PECULIARITIES OF IDENTIFICATION IN THE PRESENCE OF MODEL ERRORS

Modelling errors are often the limiting factor in identification problems. Therefore, it is important to qualify their impact on the estimated plant model parameters θˆ. This paper qualifies the influence of model errors and disturbing noise level on (i) the asymptotic value θ* (estimate for an infinite amount of data) of θˆ, and (ii) the asymptotic (amount of data going to infinity) covariance matrix Cov( θˆ) of θˆ. The theory is elaborated on a time domain and a frequency domain estimator.

Rik Pintelon | Johan Schoukens

[1] Lennart Ljung,et al. Estimating model variance in the case of undermodeling , 1992 .

[2] L. Ljung,et al. L2 Model Reduction and Variance Reduction , 2000 .

[3] PARAMETER VALUES OF ARMA MODELS MINIMISING THE ONE-STEP-AHEAD PREDICTION ERROR WHEN THE TRUE SYSTEM IS NOT IN THE MODEL SET , 1983 .

[4] L. Ljung,et al. Estimating the variance in case of undermodeling using bootstrap , 1999, Proceedings of the 38th IEEE Conference on Decision and Control (Cat. No.99CH36304).