Comprehensive Sensory Quality Evaluation of Dianhong Congou Tea Infusions Using Intelligent Sensory Multi-source Information Fusion Technology

According to the results of sensory evaluation by tea tasters, 21 Dianhong Congou tea infusion samples were divided into three categories: A, B, and C. Using multi-source information fusion based on electronic nose, electronic tongue and electronic eye, a 112-dimensional dataset was obtained to characterize the comprehensive sensory quality of tea infusions. The results showed that orthogonal partial least squares-discriminant analysis (OPLS-DA) could effectively distinguish the three types of tea infusion (R_Y² = 0.937, Q² = 0.900). Seven variables including S1 (AHS), S5 (NMS), B71 and B32 (nerol), B27 (myrcene), S4 (CTS), and B29 (limonene) were considered important in the classification model with variable importance in projection (VIP) score greater than 1.5. Furthermore, quantitative prediction of the comprehensive quality score of tea infusions was achieved using partial least squares regression (PLSR) and stepwise multivariate linear regression (SMLR). The absolute error and the relative error of the former were less than 1.86 points and 2.06%, while those of the latter were less than 3.57 points and 4.07%, respectively. These results demonstrate that multi-source information fusion combined with multivariate statistical analysis can allow for qualitative and quantitative evaluation of the comprehensive quality of Dianhong Congou tea infusions, which provides effective technical support for the quality control and evaluation of tea infusions.