Title

Identification Of Structural Changes By Using Statistical Pattern Recognition

Abstract

Statistical pattern recognition applications have gained considerable attention in to detect structural changes (i.e. damage) that occur over time. Outlier analysis of the features/metrics obtained from the structural health monitoring (SHM) data is used to detect the changes in the structural condition. One of the most challenging aspects of structural condition evaluation of existing constructed facilities by experimental data could be considered as the stochastic nature of the problem. Most of the time, the inherent uncertainties, such as due to environmental effects, can mask the evidence of structural change in the data. Detection of the structural changes is also very dependent on the redundancy of the structure and the level of damage. In this study, a two span 18 ft by 6 ft steel grid is used for different levels of damage on this redundant structure to investigate the effectiveness of statistical methods that have shown promise on other structures. Ambient vibration data is collected. This study accompanies time series modeling (Auto-Regressive model) of response data with a Mahalanobis distance based outlier detection algorithm to detect the structural changes/damage. After explaining the theoretical background of the methodology, example results coming from laboratory tests are presented in a comparative fashion.

Publication Date

1-1-2007

Publication Title

Structural Health Monitoring 2007: Quantification, Validation, and Implementation - Proceedings of the 6th International Workshop on Structural Health Monitoring, IWSHM 2007

Volume

2

Number of Pages

1332-1339

Document Type

Article; Proceedings Paper

Personal Identifier

scopus

Socpus ID

84945192519 (Scopus)

Source API URL

https://api.elsevier.com/content/abstract/scopus_id/84945192519

This document is currently not available here.

Share

COinS