Research On The Identification Of Wine Origin In Five Producing Areas Of Australia Based On Multi-fingerprint Technology

Wine is a kind of product that is beneficial to human health and has high economic value.Driven by economic interests,some unscrupulous manufacturers imitated high-quality wines or mislabeled them,disrupting the wine market economy,infringing on consumer rights,and hindering the development of the wine market.Therefore,it is particularly important to study wine producing area identification technology.This subject determined the volatile components,stable isotopes and mineral elements of wine from different producing areas,and combined them to identify the origin of wine.The aim was to obtain more wine fingerprint information and construct a more complete wine fingerprint spectrum system,which can be used for the original research on the classification,identification and protection of wines in producing areas provides scientific and reasonable technical selection reference and data support.The main research contents and results are as follows:The HS-SPME-GC-MS technology was used to determine the volatile components in wine from five areas in Australia.A total of 109 volatile components were identified;there were 45 volatile components in the five producing areas;the wines from the five producing areas had a rich variety of volatile components such as esters,alkanes and alcohols,while the higher concentration of volatile compounds were esters,alcohols and acids.Simultaneously,the unique volatile compounds in the wine samples of each producing area were determined.All these indicated that there are certain differences in the types and quality concentrations of volatile components in the five wine producing areas of Australia.The analysis of variance(ANOVA)technique was performed on the 45 volatile components in the wine samples from the five areas of Australia that have been identified,and 25 characteristic volatile compounds were screened out.Then the 25 characteristic volatile components selected were used as the input variables of the PCA and LDA statistical analysis to establish a model.Six principal components were extracted from the PCA analysis,and the cumulative variance contribution rate was 83.59%.The accuracy rates of the original classification and cross-validation classification of wine samples from the five areas of Australia achieved by LDA analysis were 60.60%and 56.40%,respectively.The EA-IRMS and EQ-IRMS were used to determine and analyze the ?13C and ?18O in wine samples from five areas in Australia.The results of ANOVA analysis showed that there was no significant difference in ?13C and significant difference in ?18O in wine from the five regions of Australia.At the same time,the box-plot showed the natural distribution of carbon and oxygen stable isotopes in wine from various producing areas:the distribution of ?13C was relatively concentrated,and the dispersion of ?18O was significantly greater than that of ?13C,but the distribution of ?18O in a single producing area was relatively concentrated.Then ?13C and ?18O were used as the input variables of PCA and LDA statistical analysis to establish the model.Only one principal component was extracted in the PCA analysis,and the cumulative variance contribution rate of this principal component was 61.82%.The accuracy rates of the original classification and cross-validation classification of wine samples from the five areas of Australia achieved by LDA analysis were both 46.80%.The ICP-MS was used to determine and analyze 31 mineral elements in wine samples from five areas in Australia.Analysis of variance screened out 21 characteristic mineral elements.Then a model was established with the 21 characteristic mineral elements selected as input variables for the statistical analysis of PCA and LDA.Seven principal components were extracted in the PCA analysis,and the cumulative variance contribution rate of these seven principal components was 75.18%.The accuracy rates of the original classification and cross-validation classification of wine samples from the five areas of Australia achieved by LDA analysis were 86.20%and 75.50%,respectively.Using different combinations of the three types of indicators as input variables for PCA and LDA statistical analysis,a multi-fingerprint technology identification model was established.The three types of indicators were the ?13C and ?18O of wine samples from the five producing areas of Australia,as well as 25 characteristic volatile compounds and 21 characteristic mineral elements that have been screened out.A model was constructed based on volatile component analysis and stable isotope technology.The PCA model extracted 7 principal components,and the cumulative variance contribution rate of the 7 principal components was 83.18%.The accuracy rates of the original classification and cross-validation classification of the LDA model were 71.30%and 59.60%,respectively.A model was constructed based on volatile components and mineral element analysis technology.The PCA model extracted 12 principal components,and the cumulative variance contribution rate of the 12 principal components was 82.39%.The accuracy rates of the original classification and cross-validation classification of the LDA model were 95.70%and 81.90%,respectively.A model was constructed based on stable isotope and mineral element analysis technology.The PCA model extracted 7 principal components,and the cumulative variance contribution rate of the 7 principal components was 71.91%.The accuracy rates of the original classification and cross-validation classification of the LDA model were 86.20%and 75.50%,respectively.A model was constructed based on volatile components and elemental analysis techniques.The PCA model extracted 13 principal components,and the cumulative variance contribution rate of the 13 principal components was 82.93%.The accuracy rates of the original classification and cross-validation classification of the LDA model were 96.80%and 85.10%,respectively.At the same time,the PCA and LDA discrimination models based on the combination of three technologies were analyzed and compared,and it was found that the PCA model based on the volatile component analysis technology had the highest cumulative variance contribution rate of the principal components extracted.The original classification and cross-validation classification accuracy rates of the LDA model constructed based on the three analysis techniques of volatile components,stable isotopes and mineral elements were the highest.