采用杭州西湖区、绍兴新昌县和丽水市辖四县区生产的三类扁形茶样本,按单一品种和混合品种两种策略,进行了基于多元化学指纹图谱和逐步判别技术的茶样产地鉴别分析方法研究。试验结果表明,无论是采用单一品种样本,还是采用混合品种样本进行判别分析,都能得到有效的判别函数,使三个不同产地的扁形茶得到有效区分。除迎霜品种有一个外部验证样本判别错误外,其它单一品种的判别分析的回判成功率和验证成功率都是100%。9个品种的混合品种判别试验中,训练集143个样本的回判正确率为93.7%,外部验证样本的判别正确率91.7%,表明采用化学指纹图谱方法结合判别技术对茶产品的产地属性进行鉴别或验证分析是可行的。
The method of Discriminanting Classification of Flatten-shaped Green Tea Production Area Based on Multiple Chemical Fingerprint was studied by way of using samples from single cultivar and from mixed cultivars, with three groups of samples produced in West Lake District Hangzhou, Xinchang County Shaoxin and Lishui respectively. The result showed that samples from three district could be distinguished effectively by using those discriminant method obtained form the investigation, whether single cultivar samples or mixed cultivar samples were used. The recognition accuracies of the training sample set and the test sample set were all 100% except one Yingshuang cultivar test sample. When using mixed samples of 9 cultivars, the recognition accuracy of the training sample set 143 samples was 93.7% and that the test sample set was 91.7%. Those above result revealed that it was possible to distinguish or verify the producing area attribute of flatten-shaped green tea products by combining the techniques of multiple chemical fingerprint and discriminant classification.
[1] 宋冠群, 董文举, 林金明, 等. 茶叶指纹谱的毛细管胶束电动色谱法研究[J]. 色谱, 2003, 2l(4): 359~362.
[2] 罗一帆, 郭振飞, 许旋, 等. 广东岭头单枞茶高效液相色谱指纹图谱的研究[J]. 食品科学, 2005, 26(4): 206~209.
[3] 陈全胜, 赵杰文, 张海东, 等. 基于支持向量机的近红外光谱鉴别茶叶的真伪[J]. 光学学报, 2006, 26(6):933-937.
[4] 张铭光, 袁敏, 袁鹏, 等. 普洱茶热脱附一裂解色谱指纹图谱研究[J]. 华南师范大学学报(自然科学版), 2006(3): 96~101.
[5] 范骁辉, 叶正良, 程翼宇. 基于信息融合的中药多元色谱指纹图谱相似性计算方法[J]. 高等学校化学学报, 2006, 27(1): 26~29.
[6] 王丽鸳, 成浩, 周健, 等. 绿茶数字化多元化学指纹图谱建立初探[J]. 茶叶科学, 2007, 27(4): 335~342.