Construction of Discriminant Models for Four Edible Animal Oils Based on Gas Chromatography-Mass Spectrometry with Chemometrics

KeyWord:animal oil  thermal desorption/gas chromatography-mass spectrometry(TD/GC-MS)  volatile components  principal component analysis  cluster analysis  discriminant analysis

Abstract:

      The volatile components of four different types of animal oils£¬eª±g.chicken oil£¬tallow£¬goat oil and lard oil£¬were determined.In perspective of the volatile flavor substances of animal oils£¬the possibilities of distinguishing and classifying different types of animal oils were explored.A gas chromatography-mass spectrometry with thermal desorption(TD/GC-MS) was performed for the determination of the quasi-determinacy and relative contents of volatile components in four animal oils£¬while principal component analysis£¬cluster analysis and discriminant analysis were applied in the analysis of the principal components of volatiles in the oils based on SPSS 22ª±0 statistical software.According to the flavor threshold value and characteristics of volatile compounds and their relative contents£¬the principal component analysis(PCA) was used after the variables and index gains and losses.A total of 19 main flavor substances were screened out from the 53 volatile components in the four animal oils£¬and their relative contents were reduced to obtain three principal components.Based on this£¬three canonical discriminant function models for animal oils were established by using the results of principal component analysis in turn for cluster analysis and discriminant analysis.The accuracies for discriminating the volatile components of chicken oil£¬tallow£¬goat oil and lard oil by the training models reached to 100%.