关键词: 乳粉, 超快速气相电子鼻, 挥发性风味成分, 线性判别分析, 线性支持向量机

Abstract: The ultra-fast gas chromatography electronic nose (UFGC E-nose) was used to analysis the volatile flavor components of 24 samples of infant, student, adult and middle-aged and elderly whole milk powder. Combined with linear discriminant analysis (LDA) and linear support vector machine (LSVM), the kinds of milk powder were classified and predicted. The results showed that 54 volatile flavor components were detected in four different age groups, among which 37 common components were propionaldehyde, acetone, hexanal and caproic acid, etc. Methyl propionate and 1-octen-3-one only existed in infant milk powder. (E,E)-2,4-Decadienal and γ-unsecalactone were the special components of student milk powder. Heptanal was only detected in adult milk powder. 2-Butanone and benzaldehyde were only found in middle-aged and elderly milk powder. The results of relative odor activity value evaluation and aroma radar map showed that 23 volatile components contributed to the aroma, 10 of which were the key aroma contributing compounds shared by the four age groups, and the common aroma of milk powder were milk flavor, vegetable flavor, fruit flavor, flower flavor, grass flavor, caramel flavor and tar flavor. In addition, the almond flavor was the characteristic aroma of infant milk powder. The sensory evaluation results showed that the flavor of student milk powder and adult milk powder were better than that of infant milk powder and middle-aged and elderly milk powder, and there were significant differences with the results of aroma radar maps. The classification and prediction accuracy of LDA and LSVM for milk powder at different ages were 93.3% and 94.2%, respectively.

Key words: milk powder, ultra-fast gas chromatography electronic nose (UFGC E-nose), volatile flavor components, linear discriminant analysis (LDA), linear support vector machine (LSVM)