基于气相-离子迁移谱的酿酒高粱早期霉变判别方法研究  

Early Mildew Discrimination Method of Sorghum Based on Gas Chromatograph-Ion Mobility Spectrometer

作　　者：王娜 谌松强[2] 聂秋宇 李子文 李宗朋 熊雅婷 齐琳 程亚娟 王健 张海红 Wang Na;Chen Songqiang;Nie Qiuyu;Li Ziwen;Li Zongpeng;Xiong Yating;Qi Lin;Cheng Yajuan;Wang Jian;Zhang Haihong(China National Research Institute of Food and Fermentation Industries ,Beijing 100015;Jiugui Liquor Co.Ltd.,Jishou 416000;School of Food Science and Engineering,Ningxia University ,Yinchuan 750021)

机构地区：[1]中国食品发酵工业研究院有限公司,北京100015 [2]酒鬼酒股份有限公司,吉首416000 [3]宁夏大学食品科学与工程学院,银川750021

出　　处：《中国粮油学报》2025年第2期177-185,共9页Journal of the Chinese Cereals and Oils Association

基　　金：国家重点研发计划项目(2022YFF0606803)。

摘　　要：高粱在储存及运输阶段极易受到霉菌的侵染,为快速无损检测出无感染症状的早期霉变高粱,避免其携带有害霉菌进入白酒酿造环节导致品质下降,采用气相色谱-离子迁移谱技术(GC-IMS)结合机器学习算法,探究高粱不同霉变程度挥发性物质差异及其贡献度,筛选能够表征不同霉变程度的特征物质并建立判别模型。结果显示,在所有高粱样品中共鉴定出38种挥发性成分,包括醛类8种,醇类6种,酮类4种,酯类6种,其他类3种;PLS-DA分析结合霉菌菌落总数可对不同霉变程度高粱进行分类,依据变量重要性投影以及最小冗余最大相关性筛选了9种关键特征性物质用于建立线性判别模型,模型准确率和精确率提高了12.5%,召回率提高了9.4%,F_(1)值提高了10.4%。该方法可用于高粱霉变程度识别。Sorghum is easily infected by fungi during storage and transportation.In order to quickly and non-destructively detect the early mildewed sorghum without obvious infection symptoms,and prevent it from carrying harmful fungi into the Baijiu-making process,which will lead to the decline of Baijiu quality,in this study,gas chromatography-ion mobility spectrometry(GC-IMS)technology was used with machine learning algorithms to explore the differences in volatile substances and their contributions at various degrees of moldiness in sorghum,aiming to screen characteristic substances that can represent different levels of mold and establish a discrimination model.The results indicated that 38 volatile components were identified from all sorghum samples,including 8 aldehydes,6 alcohols,4 ketones,6 esters and 3 others.Partial Least Squares Discriminant Analysis combined with the total mold count could classify sorghum with different degrees of moldiness.Based on the Variable Importance for the Projection and the Minimum Redundancy Maximum Relevance,9 key characteristic substances were selected to establish a linear discriminant analysis model.As a result,the model′s accuracy and precision rate increased by 12.5%,the recall rate improved by 9.4%,and the F_(1) score was enhanced by 10.4%.This method could be used for the identification of mold levels in sorghum.

关 键 词：酿酒高粱 早期霉变 气相-离子迁移谱 挥发性物质 判别模型 

分 类 号：TS261.2[轻工技术与工程—发酵工程]