Abstract: The volatile compounds of 25 brands of fragrant vinegar, rice vinegar and aged vinegar were detected by ultra-fast gas chromatography-electronic nose and were analyzed qualitatively and quantitatively by retention index and external standard method. A flavor wheel for vinegar was developed based on the odor activity values (OAV) of volatile compounds, and the volatile aroma profiles of the different kinds of vinegar were compared and their relationship with the content of volatile compounds was evaluated. Linear discriminant analysis (LDA) and random forest (RF) were used to investigate the difference in the composition of volatile compounds of different kinds of vinegar in order to distinguish them. The results showed that in total 44 volatile compounds were detected in the three kinds of vinegar, mainly including alcohols, esters, aldehydes, ketones, acids, alkanes and heterocycles, the most abundant ones being acids, followed in decreasing order by alcohols, aldehydes and alkanes. The OAVs and odor radar chart of vinegar samples showed that acetic acid, 3-methylbutyraldehyde, phenylacetaldehyde, 5-methylfurfural and guaiacol contributed a lot to the aroma. The sourness, and fruity and cocoa-like aroma of fragrant vinegar were higher than those of rice vinegar and aged vinegar, but the flowery aroma of aged vinegar was higher than that of the other two types. The results of LDA showed that there were significant differences among different brands and types of vinegar. Using random forest machine learning, the three types of vinegar were discriminated with an accuracy of 100%, and 3-methylbutanol, guaiacol, benzaldehyde and acetone were selected as important compounds to distinguish different kinds of vinegar.

Key words: vinegar; ultra-fast gas chromatography-electronic nose; volatile components; odor radar map; linear discriminant analysis; random forest