Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

General perspectives on model construction and evaluation for stochastic estimation, with application to a blunt trailing edge wake

General perspectives on model construction and evaluation for stochastic estimation, with... The technique of stochastic estimation is examined as a specific application of linear least squares modelling. Factors that are relevant to the objectives of estimation in fluids, such as the number of sensors, the use of multiple time lags, and the strength of linear correlations, are discussed in the context of a general regression formulation. We consolidate the established findings of several research fields in order to outline clearly the potential pitfalls and reasonable performance expectations of these empirical strategies. Experimental measurements of velocity and fluctuating pressure in the wake of a blunt trailing edge body are used for quantitative illustration of key considerations for model construction and performance evaluation. It is emphasized that estimator accuracy is influenced strongly by the physical relationships among the measured variables, in addition to their correlation with the estimated variable. The evaluation of several performance metrics on an independent test set provides valuable information for the selection of a suitably complex model. In particular, “variance inflation” is interpreted as an indicator of the potential amplification of noise by a stochastic estimator. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Experiments in Fluids Springer Journals

General perspectives on model construction and evaluation for stochastic estimation, with application to a blunt trailing edge wake

Experiments in Fluids , Volume 55 (7) – Jun 19, 2014

Loading next page...
1
 
/lp/springer_journal/general-perspectives-on-model-construction-and-evaluation-for-0LjjRJUh2D

References (63)

Publisher
Springer Journals
Copyright
Copyright © 2014 by Springer-Verlag Berlin Heidelberg
Subject
Engineering; Engineering Fluid Dynamics; Fluid- and Aerodynamics; Engineering Thermodynamics, Heat and Mass Transfer
ISSN
0723-4864
eISSN
1432-1114
DOI
10.1007/s00348-014-1756-y
Publisher site
See Article on Publisher Site

Abstract

The technique of stochastic estimation is examined as a specific application of linear least squares modelling. Factors that are relevant to the objectives of estimation in fluids, such as the number of sensors, the use of multiple time lags, and the strength of linear correlations, are discussed in the context of a general regression formulation. We consolidate the established findings of several research fields in order to outline clearly the potential pitfalls and reasonable performance expectations of these empirical strategies. Experimental measurements of velocity and fluctuating pressure in the wake of a blunt trailing edge body are used for quantitative illustration of key considerations for model construction and performance evaluation. It is emphasized that estimator accuracy is influenced strongly by the physical relationships among the measured variables, in addition to their correlation with the estimated variable. The evaluation of several performance metrics on an independent test set provides valuable information for the selection of a suitably complex model. In particular, “variance inflation” is interpreted as an indicator of the potential amplification of noise by a stochastic estimator.

Journal

Experiments in FluidsSpringer Journals

Published: Jun 19, 2014

There are no references for this article.