|
Activity Number:
|
334
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Physical and Engineering Sciences
|
| Abstract - #306567 |
|
Title:
|
A Statistical Method for Crack Detection in Thermal Acoustics Nondestructive Evaluation Data
|
|
Author(s):
|
Chunwang Gao*+ and William Q. Meeker, Jr.
|
|
Companies:
|
Iowa State University and Iowa State University
|
|
Address:
|
46 Schilletter Village, Apt. D, Ames, IA, 50010-8746,
|
|
Keywords:
|
nondestructive evaluation ; thermal acoustics ; probability of detection ; principal components ; robust fitting ; data clustering
|
|
Abstract:
|
A systematic statistical method based on principle components analysis is developed to automatically analyze a sequence of images (movie) generated in thermal acoustic inspections to detect component anomalies such as internal cracks. The method uses principle components regression for dimension reduction in the data processing. Robust fitting and clustering are applied on principle components regression results to reduce the effect of the noise structure in the movie data and to help in setting up the detection rule. The method gives results that are consistent with a human expert detection by visual inspection of the movie after some simple signal processing. The Probability of Detection ( POD) is calculated according to the detection rules.
|
