|
Activity Number:
|
29
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, July 29, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Physical and Engineering Sciences
|
| Abstract - #310369 |
|
Title:
|
Analysis of Probability of Detection and Probability of Alarm on Sonic IR Inspection Data Using the Mixed Effects Model and Bootstrapping
|
|
Author(s):
|
Chunwang Gao*+ and William Meeker and Donna Mayton
|
|
Companies:
|
Iowa State University and Iowa State University and SAIC
|
|
Address:
|
46 Schilletter VLG APT D, Ames, IA, 50010,
|
|
Keywords:
|
Sonic IR ; Thermal Acoustic ; Mixed Effect Model ; Bootstrapping ; Nondestructive Evaluation
|
|
Abstract:
|
Sequences of infrared images were taken by Sonic IR system on 10 cracks in 7 different samples. For each crack, different levels of vibration amplitude (3 levels), pulse length (3 levels) and trigger force (3 levels) of the Sonic IR system are arranged to study the relationship between the maximum contrast of the images and the experimental setup. Time to alarm was also recorded for those having an alarm. A quadratic random effects response surface model to the combined contrast data from the 10 cracks was fitted to provide an assessment of probability of detection as a function of the experimental variables. Another linear response surface model to the combined time-to-alarm data from the 10 blade cracks are fitted to provide an assessment of the probability of alarm. Bootstrapping is used to explore methods of quantifying statistical uncertainty in the two models.
|
