Author

Hans G. Korth

Date of Award

3-14-2014

Document Type

Thesis

Degree Name

Master of Science

Department

Department of Engineering Physics

First Advisor

Timothy W. Zens, PhD.

Abstract

Polyurethane-based aircraft coating degradation was examined using Diffuse Reflectance Infrared Fourier Transform spectroscopy (DRIFTs). Two sample sets were aged in an autoclave; each containing four samples. Sample set A contained two as-cured samples and two samples treated to a level of visual degradation. Sample set B contained one sample left as-cured and three samples of differing degradation. DRIFTs spectra were collected on sample set B and singular value decomposition (SVD) was used to reveal trends between the degradation levels. Linear discriminant analysis (LDA) was then applied to the SVD coefficients to determine the most accurate spectral band for classification of unknown degradation. The 1220-850 cm-1 band proved to be the most accurate at discerning between degradation levels with 98.3% accuracy. The approach was then applied to sample set A using one each of the as-cured and degraded samples as unknown degradation levels. The prediction accuracy of LDA was 100% for the as-cured sample, but only 28% for the visibly degraded sample in the fingerprint region. When the misclassification cost was adjusted, the accuracy improved to 78%. It is recommended that this work continue towards generating a field NDI technique.

AFIT Designator

AFIT-ENP-14-M-19

DTIC Accession Number

ADA598862

Share

COinS