Quality Evaluation Of Jasmine Tea Using NIR And Chemometrics

The evaluation of the quality of jasmine tea is still mainly depends on sensory evaluation until now, which is easily affected by subjective factors. Therefore an objective method based on chemical composition is very necessary for tea quality evaluation. In this work, concentration and content of13chemical compositions (including total tea polyphenols(TP), free animo acids(aa), caffeine, water extract, GA, GC, EGC, C, EC, EGCG, GCG, ECG, CG) of112jasmine tea samples were determined and the quality of jasmine tea samples was evaluated by sense organ.The quantitative calibration models for determining the contents of13chemical compositions were established through NIR combined with Chemometrics. The optimal parameters were as follows:For the13chemical compositions, PLS (partial least square) and3999.67-10001.1cm-1were the best calculation method and range of wave number. The optimal numbers of principal components were5,8,6,9,6,6,7,8,8,8,7,7and7. The appropriate spectral data preprocessing methods were ds2, dblg2+ncl, SNV+dbl, mf+db1g2, sa3+ncl+db1, SNV+dbl, dbl+mf, dblg2+mf, dblg2+mf, dbl, dbl, SNV+db1and db1. Q-value taking all important statistical values (such as SEE, SEP, V-bias, etc.) into account can evaluate the quality of models. The higher the Q-value is, the better the calibration model will be. Q-value was0.8571,0.8539,0.8159,0.8248,0.9063,0.8526,0.8931,0.8873,0.8731,0.8629,0.8184,0.8760and0.7702.20jasmine tea samples not included in calibration and validation sets were verified to futher evaluate these quantitative calibration models. Coefficients of correlation (R) between predicted value by NIR and the actual value were0.9615,0.9450,0.7071,0.9374,0.9842,0.8334,0.9396,0.9558,0.9333,0.8743,0.7973,0.8461and0.8232. These models could be used to measure the contents of the13chemical compositions.The results of correlativity analysis were as follows:(1)For the five sensory factors, such as appearance, liquor color, aroma, taste and infused leaf were significantly and positively correlated with each other and to the sum. The five sensory factors and total quality score were logarithm relevant to the price, and the coefficients of determination (R2) were0.855、0.667、 0.836、0.900、0.906、0.914.(2)For the six taste factors, tenderness, sweetness, freshness and mellowness were positively correlated with each other and to the scores of taste, total organoleptic quality and price, while heaviness was opposite. Thickness was positively correlated to heaviness and negatively correlated to sweetness, mellowness and price.(3)The concentrations of TP, aa, water extract, GC, EGC, C, EC, EGCG, GCG and ECG in tea liquor by brewing were negatively correlated to tenderness, sweetness, freshness mellowness, taste score, total quality score and price, while were positively correlated to heaviness and thickness (except individual components). GA was opposite.(4)The contents of TP, aa, caffeine, GA, ECG and CG by extracting were positively correlated to tenderness, sweetness, freshness, mellowness, taste score, total quality score and price, while were negatively correlated to heaviness and thickness (except individual components). Water extract, GC, EGC, C, EC, EGCG and GCG were opposite.In the brewing and extracting methods,4multiple regression equations of taste score and total quality score were established. The results of back substitution were of high correlation (R) between predicted and actual value and acceptance rate was over90%. For both of brewing and extracting methods,5factor score functions,1comprehensive evaluation function and3Bayes discriminant functions were obtained along with factor analysis and discriminant analysis for each method. Recognition accuracy of back substitution was85.7%and83.9%, meanwhile the recognition accuracy of the cross-validation was84.8%and81.3%. On the whole, the discrimination results were satisfactory.This study established a basis for evaluating qulity of jasmine tea more easily and objectively.