Application of Pattern Recognition and Fuzzy Sets to the Interpretation of Chromatograms

The described algorithms show the possibilities provided by pattern recognition methods in combination with fuzzy sets for the identification of unknown sample chromatograms. Starting with handling a training set of chromatograms, a classification vector k and an identification vector e are computed for every chromatogram within the class. These data vectors are used in combination with pattern recognition methods for the identification of sample chromatograms. In principle, the algorithms are also applicable to other nonperiodic spectra (e.g., UV-, or IR-spectra). A computer program allowing such a peakspecific comparison with interactive features is under preparation.