An application of pattern recognition and infrared spectroscopy to water analysis
A mathematical methodology is presented that shows potential for the interpretation of infrared spectra through a technique of pattern recognition. A number of water samples and simple alkanes were studied to examine the sensitivity and discriminating qualities of the methodology. With 2500 comparisons each of tap water versus tap water or distilled water versus tap water, the technique performed well in selecting the targeted reagent. The same is true for the comparisons of n-pentane, as the window compound, versus n-heptane, n-octane, n-nonane and n-decane. Comparisons of these n-alkanes to n-pentane gave fitting tolerances of 15.3, 21.2, 20.9 and 24.7%, respectively. When n-pentane was compared to itself, the tolerance fit was 2.7 %, showing the ease of discrimination. These results suggest that this computer-aided phase space transformation method is sensitive, offers good analytical precision, and is capable of detecting small differences in the infrared spectra of compounds and mixture studied. Preliminary data also suggest that the method has potential for monitoring the quality of water. © 1991, Taylor & Francis Group, LLC. All rights reserved.
Witelski, T; Ng, P; Joseph Lundy, JY; Bove, J
Volume / Issue
Start / End Page
Electronic International Standard Serial Number (EISSN)
International Standard Serial Number (ISSN)
Digital Object Identifier (DOI)