国产红葡萄酒抗氧化活性比较分析及其在品种鉴别中的应用

Comparison analysis of antioxidant activity in Chinese red wines and its application in varietal discrimination

中文关键词:  国产红葡萄酒  抗氧化活性  主成分分析  正交偏最小二乘-判别分析  支持向量机  品种识别

英文关键词:Chinese red wine  antioxidant activity  principal component analysis  orthogonal partial least squares-discriminant analysis  support vector machine  varietal identification

中文摘要:

      目的 利用不同品种国产红葡萄酒的抗氧化活性差异建立品种快速鉴别模型。方法 通过测定并比较不同品种国产红葡萄酒(黑皮诺、梅鹿辄、蛇龙珠和赤霞珠)的总酚含量、总黄酮含量、铜离子还原能力、1,1-二苯基-2-三硝基苯肼自由基和2,2’-联氮-二(3-乙基-苯并噻唑-6-磺酸)二铵盐阳离子自由基清除能力, 同时结合主成分分析(principal component analysis, PCA)和正交偏最小二乘-判别分析(orthogonal partial least squares-discriminant analysis, OPLS-DA)以及支持向量机(support vector machine, SVM)算法, 构建国产红葡萄酒品种识别模型。结果 对比分析发现4个品种的国产红葡萄酒抗氧化活性顺序为: 赤霞珠>梅鹿辄>蛇龙珠>黑皮诺。基于此, 构建的PCA (R2Xcum=0.819)以及OPLS-DA (R2Xcum=0.952、R2Ycum=0.424、Q2cum=0.387)模型能有效表征数据特征与组间差异; SVM品种识别模型中交叉验证准确率为94.40%, 训练集准确率达100.00%、测试集准确率为96.26%。结论 基于国产红葡萄酒抗氧化活性的品种特异性差异, 结合多元统计学方法及机器学习算法, 可以实现对国产红葡萄酒品种的快速、准确鉴别。

英文摘要:

      Objective To establish a rapid model for varietal discrimination of Chinese red wines by leveraging the differences in antioxidant activity among different varieties. Methods The total phenolic content, total flavonoid content, cupric ion reducing antioxidant capacity, 1,1-diphenyl-2-picryl-hydrazyl radical scavenging capacity and 2,2’-azinobis(3-ethylbenzothiazoline-6-sulfonic acid) ammonium salt radical cation scavenging capacity of different varieties of Chinese red wines (Pinot Noir, Merlot, Cabernet Gernischt and Cabernet Sauvignon) were determined and compared. Meanwhile, principal component analysis (PCA), orthogonal partial least squares-discriminant analysis (OPLS-DA) and support vector machine (SVM) algorithms were combined to establish a varietal identification model for Chinese red wines. Results Comparative analysis revealed that the order of antioxidant activity of the 4 varieties was: Cabernet Sauvignon>Merlot>Cabernet Gernischt>Pinot Noir. On this basis, the constructed PCA (R2Xcum=0.819) and OPLS-DA (R2Xcum=0.952, R2Ycum=0.424, Q2cum=0.387) models effectively characterized the data features and intergroup differences. The SVM varietal identification model achieved a cross-validation accuracy of 94.40%, with 100.00% accuracy in the training set and 96.26% accuracy in the test set. Conclusion Based on the variety-specific antioxidant activity differences of Chinese red wines, combined with multivariate statistical methods and machine learning algorithms, rapid and accurate discrimination of Chinese red wine varieties can be effectively achieved.