Abstract：Identification of green teas was made according to catechins and caffeine contents.Five varieties of green teas were tested in the experiment.First,the contents of catechins（epigallocatechin,epicatechincatechin,epigallocatechin gallate,epicatechin gallate） and caffeine were determined by HPLC.Then,Linear Discriminant Analysis（LDA）,K-Nearest Neighbors（KNN）,and Back-Propagation Artificial Neural Networks（BP-ANN） were applied comparatively to identify tea varieties as pattern recognition method.Experimental results showed that the best identification model was achieved by BP-ANN,and the identification rates for calibration set and prediction set samples were 98%and 92%,respectively.It is feasible to identify green tea varieties according to the analysis of catechins and caffeine contents by HPLC coupled with ANN pattern recognition.
2010, Vol. 31 
