湖南农业大学学报（自然科学版）2018，Vol.44Issue(3)：256-260,5.DOI:10.13331/j.cnki.jhau.2018.03.006
基于逐步判别分析的卷烟叶组配方的定向模块设计
Oriented module design for cigarette blend based on stepwise discriminant analysis
摘要
Abstract
To guide the oriented design of the functional module in a certain cigarette brand from China Tobacco Jiangsu Industrial Co. Ltd., sensory analysis and chemical analysis were applied on the 236 tobacco samples from the storage.According to the use character, 4 functional groups including the group for increasing the quality and quantity of the aroma, the group for enhancing the concentration and strength, the group for balancing the cigarette smoke and the group for filling were classified based the 216 tobacco samples already used in the cigarette blend. Ninety-six representative tobacco samples from these 4 functional groups were selected, and the remaining 120 tobacco samples were optimized by using K-Nearest Neighbor method to make sure all the 216 tobacco samples can be divided into the 4 functional groups.Stepwise discriminant analysis was applied to establish the functional recognition model. The results showed that sevenindexes including density of smoke, quantity of aroma, diffusiveness, softness, dryness, aftertaste and total alkaloids were used in the recognition functions and the correct rate of self-verification method and leave one interaction test methodwere 93.1％ and 92.6％, respectively. Twenty new tobacco samples tend to be used in the cigarette blend were predicted by the functional recognition model, and the discriminant accuracy reached 100％.关键词
卷烟/功能性识别模型/逐步判别分析/定向模块设计/感官质量/化学成分Key words
cigarette/functional recognition model/stepwise discriminant analysis/oriented module design/sensory quality/chemical components分类
轻工纺织引用本文复制引用
张天兵,付金存,胡宗玉,张建彭,孙海平,叶远青,陈晶波..基于逐步判别分析的卷烟叶组配方的定向模块设计[J].湖南农业大学学报（自然科学版）,2018,44(3):256-260,5.基金项目
江苏中烟工业有限责任公司项目(201809) （201809）
