基于因子分析和神经网络分析预测评价酱香酒轮次与工艺  

Predictive evaluation of the batch and technology of sauce-flavor Baijiu based on factor analysis and neural network analysis

作　　者：何菲 陈仁远[2] HE Fei;CHEN Renyuan(Renhuai Research Institute of Sauce-flavor Baijiu,Renhuai 564500,China;Renhuai Branch of Guizhou Provincial Product Quality Supervision and Inspection Institute,Renhuai 564500 China)

机构地区：[1]仁怀酱香白酒科研所,贵州仁怀564500 [2]贵州省产品质量监督检验院仁怀分院,贵州仁怀564500

出　　处：《中国酿造》2020年第2期162-168,共7页China Brewing

基　　金：贵州仁怀地方科研课题专项基金项目(J020180002)

摘　　要：以酱香轮次基酒和酱香成品酒为研究对象,采用气相色谱-质谱联用(GC-MS)法测定其58组风味物质计算出气味活力值,并进行了因子分析和神经网络分析。结果表明,多层感知器对仁怀大曲酱香一至七轮次基酒的轮次预测正确百分比达80.2%,酱香成品酒的酱香工艺预测正确百分比达94.5%以上。通过多层感知器自变量重要性与因子分析成分矩阵共同分析发现,酱香基酒轮次的预测与异戊醛、乙酸、苯乙酸乙酯、糠醛、乙酸乙酯、2,3,5-三甲基吡嗪等化合物存在较大的相关性;酱香成品酒工艺的预测与辛酸乙酯、正己醇、癸酸乙酯、2,3,5-三甲基吡嗪、辛酸、丁酸等化合物存在较大的相关性。Using sauce-flavor batch base liquor and finished Baijiu(Chinese liquor)as research object,the odor activity value of 58 groups of flavor substances determined by GC-MS was calculated,and the data were analyzed by factors analysis and neural network analysis.The results of multilayer sensor analysis showed that the correct percentage of the prediction of the first to seventh batch sauce-flavor base liquor of Renhuai Daqu was 80.2%,and the correct percentage of the prediction of sauce-flavor finished Baijiu was more than 94.5%.Through the importance of multilayer perceptron independent variable and factor analysis component matrix study,results showed that there was greater correlation between the forecast of the batch of sauce-flavor base liquor and the content of isovaleraldehyde,acetic acid,ethyl phenyl acetate,furfural,ethyl acetate,2,3,5-trimethyl pyrazine and other compounds.The technological prediction of sauce-flavor finished Baijiu was greatly correlated with ethyl octanoic,n-hexanol,ethyl decanoate,2,3,5-trimethyl pyrazine,octanoic acid,butyric acid and other compounds.

关 键 词：酱香型白酒 酱香轮次基酒 酱香白酒工艺 因子分析 神经网络分析 

分 类 号：TS262.3[轻工技术与工程—发酵工程]