Degree Type

Thesis

Date of Award

1995

Degree Name

Master of Science

Department

Electrical and Computer Engineering

Abstract

One important application of nondestructive evaluation (NDE) methods involves the classification of the types of defects present in materials based on the measured response of those defects to different types of interrogating energy. This classification process typically involves 1) the transformation of the "raw" NDE measurements into domains more suitable for classification purposes, 2) the extraction of features in those domains that are distinguishing characteristics of different flaw types, and 3) the application of a particular classification method for making a decision on the type of flaw present (based on the strength of the features). Here, we describe the elements of a software system, CLASS, that performs the entire classification process for those NDE methods where the underlying NDE measurements are essentially one-dimensional (such as ultrasonics). This research effort is concerned with the design and implementation of code which comprises the CLASS software package. Some of the important features of CLASS are described, including its extensive use of a user-friendly graphical interface, its use of classification algorithms that are primarily data driven (i.e. they require little user expertise in building the classifiers) and the ability to easily extend CLASS by adding new features and classification methods.

DOI

https://doi.org/10.31274/rtd-180813-6078

Publisher

Digital Repository @ Iowa State University, http://lib.dr.iastate.edu

Copyright Owner

Loren Gene Knutson

Language

en

Date Available

2013-12-10

File Format

application/pdf

File Size

101 pages

Share

COinS