黄金茶绿茶风味轮及滋味分类模型的构建

Construction of flavor wheel and taste classification model forHuangjincha green tea

为探究以黄金茶理化成分对滋味类型判别的可靠性,对52个湖南湘西州黄金茶绿茶进行了感官审评和理化成分检测,并构建了黄金茶绿茶风味轮和滋味类型贝叶斯(Bayes)预测模型。结果表明,感官审评可将黄金茶分为醇厚型、鲜爽型和其他型3种滋味类型;同时以茶叶理化成分为基础,主成分分析和二阶聚类分析的结果与茶叶感官审评分类结果一致;预测模型对滋味分类具有较高的准确性,其自身验证的正确率为100%,交叉验证的正确率为89.74%,外部验证正确率为92.31%,可用于黄金茶绿茶滋味类型的判别分析。该研究为黄金茶绿茶提供了一种有效的滋味类型量化方法。

To investigate the reliability of the physical and chemical composition of Huangjincha to discriminate the taste types,sensory review and physical and chemical composition testing was conducted on 52 Huangjincha green teas in Xiangxi,Hunan province,and a Bayesian prediction model of Huangjincha green tea flavor wheel and taste types was constructed.Results showed that the sensory review could classify golden tea into three taste types:mellow,fresh,and other;meanwhile,the results of principal component analysis and second-order cluster analysis were consistent with the classification results of the tea sensory review based on the physical and chemical components of tea.The prediction model had high accuracy for taste classification,with the accuracy of its validation being 100%,the correct rate of cross-validation being 89.74,and the correct rate of external validation The accuracy of the prediction model was 100%for its validation,89.74%for cross-validation,and 92.31%for external validation.This study provides an effective method to quantify the taste type of Huangjincha green tea.