PARAMETER SELECTION IN INDIRECT MEASUREMENTS

A. G. Polak, J. Mroczka

Abstract

An alternative method of dealing with ill-conditioned problems in model identification is introduced. It consists in estimation of parameters selected on the ground of the sensitivity analysis and fixing remaining ones as constants. Statistical analysis of the resulting estimator in terms of estimate bias and variance is presented together with a realistic simulation example of the maximum expiration model. Results of simulation show that selection of parameters improves numerical conditioning of model identification and decreases variance of estimates.

Keywords

indirect measurement, ill-conditioned problem, parameter selection

Download

IMEKO-WC-2000-TC7-P203.pdf

DOI

-

IMEKO TC

TC7 - Measurement Science

Back to the proceedings