Bias, variance and optimal experiment design: Some comments on closed loop identification (Q4524141)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 1551942
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | Bias, variance and optimal experiment design: Some comments on closed loop identification |
scientific article; zbMATH DE number 1551942 |
Statements
5 November 2001
0 references
bias
0 references
variance
0 references
prediction error estimates
0 references
closed loop identification
0 references
linear discrete-time systems
0 references
data spectrum
0 references
identifiability
0 references
constraints on the output power
0 references
optimal regulator
0 references
0.8238338828086853
0 references
0.8037625551223755
0 references
0.803135871887207
0 references
0.7849686741828918
0 references
0.7847886085510254
0 references
Bias, variance and optimal experiment design: Some comments on closed loop identification (English)
0 references
Some basic results concerning bias and variance of prediction error estimates in closed loop identification of linear discrete-time systems have been re-examined in the common framework of the data spectrum. Direct insights into the identifiability question, bias and variance of the estimated parameters have been given. It was shown that the indirect identification method yields consistent estimates of the dynamics, even under an erroneous noise model. Also the optimum experiment design problem, aimed at minimizing the covariance of the estimated dynamics under various realistic constraints, has been discussed. It was evidenced that closed loop experiments are optimal for any criteria that involve any kind of constraints on the output power. It was also observed that the optimal regulator in the identification experiment can be found by solving the discrete-time LQ problem and that it does not depend on the weighting function in the design criterion. For the entire collection see [Zbl 0937.93004].
0 references