Data compression of UV spectra for library search and rapid simultaneous determinations of mixture in flowing stream analysis

Abstract

The purpose of the present investigation is concerned with the efficient use of the qualitative information contained in UV spectra for the streaming analysis. Two hundred fifty six serial readouts from the linear diode array were transformed by fast Hadamard transformation (0.056 s) and the 256 resultant terms were compressed by discarding their higher sequency terms. To find the compression ratio[=(256 minus numbers of discarded terms)/256] characteristic of UV spectra, the library of spectra of all 228 monocyclic aromatic compounds from UV ATLAS were constructed and searched with the normalized terms of a sample comperessed with various compression ratio. It was found that the use of the first 20~40 terms out of 256 trnsformed terms, with compression ratio of 0.08~0.16, gave good cognition rate. Real time simultaneous determinations of the multicomponent solution in streaming analysis were satisfactorily carried out by calculating inner-product (0.076 s by LSI-11 microcomputer) of compressed unnormalized data obtained from the mixed solution and cognition vector of each component predetermined by using a set of solutions, of known compositions. © 1988, The Japan Society for Analytical Chemistry. All rights reserved.