| Chinese tea, originate in China, which has a wide variety. According to the traditional processing methods, it can be divided into six categories respectively:green tea, black tea, white tea, oolong tea, dark tea, and yellow tea. Internationally, green tea and black tea have standards of classification. However, the other teas are lack of quantitative criteria. This phenomenon is unfavorable for the international trade of tea. In this paper, to construct the chemical classification of tea, I set up quantitative discrimination models for the other four tea types, according to their chemical component.This study collected 324 samples from the four teas across the country (white tea, oolong tea, dark tea, and yellow tea), which are different from grades, years to types. Based on the content of main chemical compositions (total caffeine, catechins, polyphenols, total free amino acids, water extract, etc.), I extracted the principal components with the data analysis software SPSS. Then, I carried on the Bayes and Fisher discrimination to establish the distinction equation. By contrastive analysis, we got the optimal discrimination model. The results are as the following:1. Select two kinds of extraction methods for the main composition factors. One is the direct use of principle component extraction. The other one is test the discovery of the water. Carried on the Bayes discriminant with two ingredient factors used at the same time. On the basis of discriminate accuracy, I find the component and factor discrimination obtained in the second way is better. composition factor is the total amount of catechins, caffeine and EGCG/total amount of catechin/total amount of free amino acid, total amount of catechins.2. Get the model of the Bayes discriminant equations:Green tea= 5.938*caff-142*Total catechins+90.261*EGCG/Total catechins+4.389*Total catechins/Total free amino acid-41.365Black tea=12.035*caff-1.169*total catechins+21.789 *EGCG/total catechins+3.576*total catechins/total free amino acid-25.848White tea=8.507 *caff-1.249*total catechins+102.472* EGCG/total catechins/+3.599*total catechins/total free amino acid-46.243Oolong tea= 5.388*caff-1.574*total catechins+99.608*EGCG/total catechins+9.664*total catechins/total free amino acid-48.890 The Bayes discriminant cross discriminant of four types of tea to achieve a discrimination rate of 94.1%.3. Fisher discriminant model of four types of tea Fisher1=0.176* total catechins+0.739*EGCG/total catechins+0.485*catechins/free amino acid-0.546*caff Fisher2=0.761*total catechins+0.343*EGCG/total catechins-0.927*catechins/free amino acid-0.052*caff Fisher3=-1.509*total catechins+0.739*EGCG/total catechins+0.173*catechins/free amino acid+0.297*caff4. Four types of tea based on the idea of gradual separation to improve the rate of discrimination and got the following models: Black tea:Fisher=0.307*total catechins-0.577*caff+0.131*total catechins/free amino acid+0.974*EGCG/total catechins Oolong tea:Fisher=-0.922*total catechins-0.214*caff+1.231*tatal catechins/free amino acid+0.244*EGCG/total catechins White tea:Fisher=-1.309*total catechins+1.121*caff+0.130*total catechins/free amino acid+0.068*EGCG/catechinsThe accuracy rates of each step in this method are 99.1%,95.7%,96.9%. Respectively, the accuracy rates are very high with less than 2 nonjudgemental number. The rate of discrimination is very high with this method which can be regarded as the optimal discriminant model for existing. |
PDF Full Text Request