An Assessment of Data Quality for Structural Identification
 

Sara Wadia-Fascetti
    Assistant Professor, Northeastern University

Francisco Rivero
    Graduate Student, Northeastern University

Masoud Sanayei
    Associate Professor, Tufts University

 

ABSTRACT

Structural identification is a popular method for nondestructive condition assessment of existing structural facilities.  Two significant error sources in the practical application of structural identification are uncertainties in measured data and modeling error.  Uncertainties in the experimental data and the overall quality of the data are addressed in this paper.  It is widely accepted that noisy and inaccurate measurements can lead to inaccuracies in the identified parameters and possibly divergence.  A set of quality indices is pro-posed to provide a consistent and quantifiable measure of data quality.  The indices are defined to represent the physical characteristics of the structural system and to highlight data qualities that are significant in structural identification.  The consistent measure makes it possible to compare different data channels and test scenarios.  The data quality indices presented in this paper are applied to experimental test data obtained from a laboratory grid structure.  Quality indices from different sets of test data are used as a rational basis for the comparison.
 

Contact Information:

Prof. S. Wadia-Fascetti (swf@neu.edu)
Dept. of Civil & Env. Engineering
Northeastern University
Boston, MA  02115

Reference:

Wadia-Fascetti, S., Rivero†, F., Sanayei, M. (submitted 5/24/99)  “An Assessment of Data Quality for Structural Identification.”  International Conference on Applications of Statistics and Probability, December 13 – 15, 1999.  Australia.
 

Back to: publications page
Back to: group home page