Abstract: The major purpose of this study was to evaluate the sweet taste quality of Congou black tea objectively, rapidly and accurately. According to sensory evaluation data obtained from experts, Congou black tea samples were divided into three categories, i.e., sweet, pure and with off-flavor. Next, the taste quality of tea infusions was comprehensively evaluated by using electronic tongue in combination with multivariate statistical analyses such as principal component analysis (PCA), partial least squares discriminant analysis (PLS-DA) and one-way analysis of variance (ANOVA). A model was established for the discrimination of sweet taste from non-sweet taste based on analysis of electronic tongue data by using binary logistic regression. Finally, the obtained results were preliminarily validated using a separate set of black tea samples. Results: PCA and PLS-DA based on the responses of electronic tongue sensors provided clear discrimination between sweet and non-sweet (pure and with off-flavor) black tea samples. The key electronic tongue sensor for taste classification was identified as the sweet taste sensor SWS (P < 0.001, by one-way ANOVA). A predictive equation for the sweet taste was established based on binary logistic regression, with an overall predictive accuracy of 93.8%. Furthermore, we confirmed the effectiveness of electronic tongue analysis by applying it to the tea samples in the validation set. In summary, the electronic tongue is able to rapidly and accurately evaluate the sweet taste of Congou black tea, and it provides a theoretical foundation for further studies of key taste components in tea, and also offers a novel tool for objective and intelligent evaluation of tea taste.

Key words: Congou black tea, electronic tongue, sweet, partial least squares discriminant analysis, binary logistic regression analysis