龙井茶叶品质的电子鼻检测方法  

Measurement of the LongJing Tea Quality by Using an Electronic Nose

作　　者：于慧春[1] 王俊[1] 张红梅[1] 于勇[1] 

机构地区：[1]浙江大学生物系统工程与食品科学学院

出　　处：《农业机械学报》2007年第7期103-106,共4页Transactions of the Chinese Society for Agricultural Machinery

基　　金：国家自然科学基金资助项目(项目编号:30571076);国家教育部新世纪优秀人才支持计划资助项目(项目编号:NCET-04-0544)

摘　　要：针对茶叶品质感官审评的不足,采用电子鼻检测手段,对4个不同等级的龙井茶作等级判别。对传感器信号进行多因素方差分析得出:不同容器容积和不同采样时刻对传感器的响应信号有着显著的影响。通过主成分(PCA)、线性判别(LDA)和BP神经网络方法对各茶叶样品进行了分类判别。PCA对于等级差别较近的茶叶区分结果不太理想;而LDA相对于PCA有较好的区分效果;设计BP神经网络拓扑结构为30-12-4,通过对网络进行适当训练,总的测试回判率可达到90%。An investigation was made to determine the four tea samples with different quality grade by using an electronic nose （e-nose）. The response signals of e-nose were analyzed under different sampling conditions by variance analysis and multivariance analysis. Analytical results showed that the different volume of vials and the different collection times have significant effect on the response signals of the e-nose. Then the data were processed using principal components analysis （PCA）, linear discrimination analysis （LDA） and artificial neural ntwork （ANN）. The results analyzed by LDA were superior to that by PCA, which could distinguish all the tea samples completely. However, PCA method could not estimate sample of A280 correctly. Further 900% correct classification was achieved for all the tea samples using the BP neural network.

关 键 词：龙井茶 电子鼻 BP神经网络 

分 类 号：TS272.7[农业科学—茶叶生产加工]