Title

Application Of Pattern Recognition Techniques To Identify Structural Change In A Laboratory Specimen

Keywords

AR-ARX modeling; Dynamic testing; Experimental modal analysis; Outlier detection; Statistical pattern recognition; Structural health monitoring; Time series analysis

Abstract

Identification of damage in a structure, or structural change in general, has been a challenging problem for the researchers in Structural Health Monitoring (SHM) area. Over the last a few decades, a number of experimental and analytical techniques have been developed and used to solve such problem. It has been has been recently accepted in the literature that the process of damage identification problem is one where statistical pattern recognition techniques can be of use because of the inherent uncertainties of the problem. Time series analysis is one of the methods, which is implemented in statistical pattern recognition applications to SHM. In previous studies, Auto-Regressive (AR) models are highly utilized for this purpose. In this study, AR model coefficients are used with different outlier detection and clustering algorithms to detect the change in the boundary conditions of a steel beam. A number of different boundary conditions are realized by using different types and amounts of elastomeric pads. The advantages and the shortcomings of the methodology are discussed in detail based on the experimental results in terms of the ability of it to detect the structural changes and localize them.

Publication Date

11-1-2007

Publication Title

Proceedings of SPIE - The International Society for Optical Engineering

Volume

6529 PART 1

Number of Pages

-

Document Type

Article; Proceedings Paper

Personal Identifier

scopus

DOI Link

https://doi.org/10.1117/12.717155

Socpus ID

35548929395 (Scopus)

Source API URL

https://api.elsevier.com/content/abstract/scopus_id/35548929395

This document is currently not available here.

Share

COinS