为实现黄酒中挥发性风味物质的快速无损检测，本研究采用气相色谱-离子迁移谱（Gas chromatography-ion mobility spectrometry，GC-IMS）联用技术结合化学计量学方法对不同酒龄黄酒中的挥发性风味物质进行了研究。黄酒样本的GC-IMS图片库显示，不同酒龄黄酒的挥发性风味物质浓度存在显著差异。采用差谱法从谱图中筛选出33个特征峰，利用其中13个数据库可查询到对应物质的特征峰进行黄酒的风味成分分析。以33个特征峰峰高作为变量，通过主成分分析（principal component analysis，PCA）进行降维，前6个主成分累计贡献率为95%，可以有效区分各组样本。分别采用线性判别（Linear Discriminant Analysis，LDA）、K-最近邻分类算法（K-Nearest Neighbor，KNN）建立酒龄判别模型。结果显示，LDA方法得到的训练集和预测集识别率分别为95％和90%，KNN的判别效果较好，训练集和预测集的识别率均达到100%。这说明，GC-IMS可以有效应用于黄酒挥发性风味物质的检测，在食品风味物质分析等领域具有广阔的应用前景。

Gas chromatography-ion mobility spectrometry (GC-IMS) combined with chemometric methods was employed as a rapid, non-destructive method for the detection of volatile flavors in Chinese rice wine. In this study, volatile organic flavor compounds in Chinese rice wine of different ages were studied. The GC-IMS image library of the Chinese rice wine samples showed significant differences in the concentration of volatile flavor substances in Chinese rice wines of different ages. Thirty-three (33) differential peaks were selected from the spectra by differential spectrum method, and thirteen (13) of them that could be queried from the databases were employed for flavor component analysis. Dimensionality reduction was achieved by principal component analysis (PCA), with 33 peaks selected and used as variables. The first 6 principal components explained 95% of the variance and the samples in each group could be distinguished. A discriminant model was established by linear discriminant analysis (LDA) and K-Nearest Neighbor (KNN). The results showed that the recognition rate of LDA was 95% and 90% for training and prediction sets respectively. The discriminant effect of KNN was better, achieving a recognition rate of 100% for both training set and prediction set. This indicates that GC-IMS can be effectively applied to the detection of volatile flavor substances in Chinese rice wine, and has broad application prospects in the field of food flavor analysis.

国家重点研发计划项目（2017YFD0400102）