摘要
为利用计算机嗅觉系统(电子鼻)快速、简便地识别同种品牌不同等级的卷烟,选取三种不同等级的"双喜"牌卷烟作为实验对象,利用PEN3电子鼻分别检测整盒(未开封)、滤嘴、烟丝、烟纸、烟气等5个方面的气味,并利用主成分分析(PCA)和主成分分析+线性判别分析(PCA+LDA)分别对该5种气味信息进行分析。最后利用相关性分析对整盒气味信息的待测样本进行了区分测试。结果显示:单独利用5种气味的信息都区分出三种等级,其中区分效果由优到劣依次是滤嘴、烟气、整盒、烟丝、烟纸。利用相关性分析方法对整盒待测样品的测试正确率达100%。
In order to fast,easily discriminate the same brand of cigarette with different levels using electronic nose,the "ShuangXi" cigarette with three different levels were used in this experiment. The odors of the entire box ( unopened) ,the ftlter,the cut tobacco,the cigarette paper,and the smoke were respectively tested by PEN3 electronic nose,and the method of PCA and the method of PCA +LDA were used to analyze the information of the five different odors. Finally ,experimental samples were tested based on the odor information of the entire box using correlation analysis. The results showed that the three levels could be discriminated from the information of the five different odors respectively, with the effects from superior to inferior followed by the filter, the smoke,the entire box, the cut tobacco, the cigarette paper. The accuracy of testing the entire box .samples by correlation analysis reached 100%.
出处
《计算机技术与发展》
2011年第11期177-181,共5页
Computer Technology and Development
基金
国家自然科学基金(60971105)
关键词
电子鼻
卷烟
主成分分析
线性判别分析
相关性分析
electronic nose
cigarette
principal component analysis
linear discriminant analysis
correlation analysis