食品与发酵工业2026，Vol.52Issue(18)：322-329,8.DOI:10.13995/j.cnki.11-1802/ts.045006
基于嗅觉诱发脑电结合提升机决策树模型的白酒典型香气喜好度研究
Study on preference of typical Baijiu aromas based on olfactory-evoked electroencephalography combined with extreme gradient boosting model
曹念 1熊小淋 1杨玉波 1黄剑炜 2李永素 1刘玄 1郝莉 2杨帆1
作者信息
- 1. 贵州茅台酒股份有限公司,贵州仁怀,564500
- 2. 杭州气味王国科技有限公司,浙江 杭州,310000
- 折叠
摘要
Abstract
The aroma of Baijiu plays a significant role in shaping consumers' pre-drinking comfort experience.To objectively evaluate how consumers perceive typical Baijiu aromas and the emotional responses these aromas elicit,a novel preference prediction method was proposed,which combines olfactory-evoked electroencephalography(EEG)with an extreme gradient boosting(XGBoost)model.EEG sig-nals were collected from 30 participants exposed to 20 typical Baijiu aromas.Then,power features from the δ,θ,α,β,and γ frequency bands were extracted to construct five biomarkers,including relaxation,concentration,alertness,acuity,and sensitivity.Subsequently,an XGBoost-based prediction model was developed.The model demonstrated robust predictive performance,with estimation errors remai-ning below 10％when compared against sensory evaluation ratings.Analysis of consumer preferences indicated that fruit-derived aromas(e.g.,apple,peach)were the most preferred,whereas unpleasant odors(e.g.,animal,fungal)received the lowest ratings.It was re-vealed from differential and correlation analyses that pleasant aromas were associated with a significant increase in feelings of relaxation and a decrease in alertness,thereby fostering a more emotionally comfortable experience.In contrast,unpleasant odors tended to heighten tension but simultaneously enhanced cognitive information processing efficiency by increasing levels of acuity.This study established a no-vel and objective methodology for evaluating Baijiu aromas,providing valuable empirical insights to inform product development and strate-gies aimed at optimizing the overall consumer sensory experience.关键词
脑电分析/白酒香气感知/喜好度/机器学习/提升机决策树Key words
electroencephalography/perception of Baijiu aroma/preference/machine learning/extreme gradient boosting引用本文复制引用
曹念,熊小淋,杨玉波,黄剑炜,李永素,刘玄,郝莉,杨帆..基于嗅觉诱发脑电结合提升机决策树模型的白酒典型香气喜好度研究[J].食品与发酵工业,2026,52(18):322-329,8.
