摘要
为了区分烤烟上部叶的不同香型风格特征,为其判断提供数学模型,以中国11个主要产烟省的65个B2F烟叶为材料,分析比较烤烟的化学成分和致香物质含量,以67种化学成分和致香物质为指标,应用逐步判别分析法对不同香型烤烟的上部叶样品进行判别,并建立Fisher判别函数。结果表明,中间香型烤烟的致香物质总量最高,清香型其次,浓香型最低,3种香型烤烟中含量最高的致香物质种类差异很小,清香型烤烟的总糖和还原糖含量极显著高于另2个香型,主导烤烟香型的化学成分和致香物质有所不同,且同一地区烤烟的香型具有一定的多样性和差异性;判别分析发现,共有12个指标(总糖、还原糖、3-甲基-1-丁醇、2-吡啶甲醛、2,4-庚二烯醛、2-乙基吡啶、壬醛、苯乙醇、苯并[c]噻酚、胡薄荷酮、丁基化羟基甲苯、3-氧代-α-紫罗兰醇)进入判别函数,并建立判别模型;采用自身验证法和交互验证法对原样品进行回判,回判正确率分别为100%、98.33%;采用该模型对19个验证样品分别进行预测,3种香型烤烟的判别正确率分别为100%、100%和83.33%。
In order to distinguish characteristics styles of aroma types of flue-cured tobacco upper leaves and establish mathematical model for judgment of aroma types, contents of chemical constituents and aroma components in flue-cured tobacco were analyzed which were based on 65 samples （B2F） from 11 main tobacco production provinces of China. Stepwise discriminant analysis was applied to discriminate tobacco leaves based on 67 chemical constituents and aroma components and the discriminant function was established. The results showed that total contents of aroma components in middle aroma type were the highest, followed by clean aroma type and then the full aroma type. There was little difference among the highest contents of aroma components in the three aroma types. And contents of total sugar and reducing sugar in clean aroma type were very significantly higher than those in the other two aroma types. Difference of chemical constituents and aroma components exited in certain dominated aroma type. Besides, there were diversity and difference in the aroma type of flue-cured tobacco of the same region. The results of discriminant analysis indicated that 12 neutral aroma components （total sugar, reducing sugar, 3-methyl-l-butanol, 2-pyridineearboxaldehyde, 2,4-heptadienal, 2-ethylpyridine, nonanal, phenethyl alcohol, benzo[c]thien, menthone, butylated hydroxytoluene, 3-oxo-α-ionol） were used in the discriminant function and found discriminant model. Self-validation method and Cross-validation method were applied to reevaluate the original sample. The accuracy of reevaluation was 100% and 98.33%, respectively. The other 19 samples were predicted by this model, the accuracy of predicted samples for the 3 kinds of aroma types （clean aroma type, middle aroma type, full aroma type） were 100%, 100%, 83.33%, respectively.
出处
《湖南农业大学学报（自然科学版）》
CAS
CSCD
北大核心
2013年第3期232-241,共10页
Journal of Hunan Agricultural University(Natural Sciences)
基金
国家烟草专卖局项目(3300806156
30300318)
关键词
烤烟
上部叶
化学成分
致香物质
香型
判别分析
flue-cured tobacco
upper leaf
chemical constituents
aroma components
aroma types
discriminant analysis
