Share Email Print
cover

Proceedings Paper

Vibration-based structural health monitoring using adaptive statistical method under varying environmental condition
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

It is well known that the dynamic properties of a structure such as natural frequencies depend not only on damage but also on environmental condition (e.g., temperature). The variation in dynamic characteristics of a structure due to environmental condition may mask damage of the structure. Without taking the change of environmental condition into account, false-positive or false-negative damage diagnosis may occur so that structural health monitoring becomes unreliable. In order to address this problem, an approach to construct a regression model based on structural responses considering environmental factors has been usually used by many researchers. The key to success of this approach is the formulation between the input and output variables of the regression model to take into account the environmental variations. However, it is quite challenging to determine proper environmental variables and measurement locations in advance for fully representing the relationship between the structural responses and the environmental variations. One alternative (i.e., novelty detection) is to remove the variations caused by environmental factors from the structural responses by using multivariate statistical analysis (e.g., principal component analysis (PCA), factor analysis, etc.). The success of this method is deeply depending on the accuracy of the description of normal condition. Generally, there is no prior information on normal condition during data acquisition, so that the normal condition is determined by subjective perspective with human-intervention. The proposed method is a novel adaptive multivariate statistical analysis for monitoring of structural damage detection under environmental change. One advantage of this method is the ability of a generative learning to capture the intrinsic characteristics of the normal condition. The proposed method is tested on numerically simulated data for a range of noise in measurement under environmental variation. A comparative study with conventional methods (i.e., fixed reference scheme) demonstrates the superior performance of the proposed method for structural damage detection.

Paper Details

Date Published: 9 March 2014
PDF: 9 pages
Proc. SPIE 9064, Health Monitoring of Structural and Biological Systems 2014, 90640T (9 March 2014); doi: 10.1117/12.2046088
Show Author Affiliations
Seung-Seop Jin, KAIST (Korea, Republic of)
Hyung-Jo Jung, KAIST (Korea, Republic of)


Published in SPIE Proceedings Vol. 9064:
Health Monitoring of Structural and Biological Systems 2014
Tribikram Kundu, Editor(s)

© SPIE. Terms of Use
Back to Top