关键词: 酱香白酒尾酒；挥发性有机化合物；质量分级；机器学习；标志化合物

Abstract: In this study, the major volatile substances and potential compounds in Jiangxiangxing Baijiu tails were analyzed by headspace solid-phase microextraction (HS-SPME) coupled with gas chromatography-mass spectrometry (GC-MS) and gas chromatography with flame ionization detector (GC-FID), aiming at providing a scientific basis for the multilevel management and efficient utilization of Jiangxiangxing Baijiu tails. The results indicated that at different distillation stages, esters, acids, alcohols, and aldehydes accounted for over 42.98%, 22.62%, 4.64%, and 2.73% of the total volatile organic compounds, respectively. At each stage, these four classes collectively constituted over 96.28% of the total volatile organic compounds detected. As the distillation process proceeded, the proportions of esters and alcohols gradually decreased, while the proportions of acids and aldehydes gradually increased. The combination of cluster analysis and machine learning identified 2-pentanone, neopentanol, hexyl acetate, ethyl oleate, ethyl linoleate ethyl palmitate as the chemical markers to distinguish between different grades of flavor structure of Jiangxiangxing Baijiu tails during distillation. The results of cluster analysis and ethanol kinetic modeling indicated that the distillation fraction collected in the first 2 minutes or with an ethanol volume fraction above 42.06% was considered as high quality, that collected between 3 and 7 minutes or with ethanol volume fractions ranging from 23.17% to 42.06% as medium quality, and that collected from 8 min onward or with ethanol volume fractions below 23.17% as low quality.

Key words: Jiangxiangxing Baijiu tails; volatile organic compounds; quality grading; machine learning; marker compounds