ABSTRACT Adulteration of honey is a major problem in the food industry. The purpose of the present study was to classify different types of monofloral honey based on physicochemical characterization and analysis of phenolic compounds coupled with chemometrics methods. The methods for classification were trialed on a wide range of honey samples from different floral origins. For thyme, jujube, coriander, barberry, acacia and alfalfa honey samples, principal component analysis combined with discriminant analysis (PCA-DA) and partial least squares combined with discriminant analysis (PLS-DA) were trialed. The results indicate that the botanical origin of the honey affects the profile of flavonoids and phenolic compounds. For example, jujube honey samples had the highest amounts of hesperetin and chrysin, while thyme honey had the maximum amount of caffeic acid; the highest levels of quercetin and p-coumaric acid were found in coriander honey. To reduce the numbers of independent variables for modeling, the principal component analysis (PCA) algorithm was used. The three scores extracted from PCA had 83.17% variance. The classification results show that PLS-DA was successfully used to predict the class membership of honey samples (100%), but PCA-DA gave the lowest correct classification rate (97%).

中文翻译：

---

根据蜂蜜的酚类成分和理化特性结合化学计量学确定蜂蜜的花​​来源

摘要 蜂蜜掺假是食品工业中的一个主要问题。本研究的目的是基于物理化学表征和酚类化合物的分析以及化学计量学方法对不同类型的单花蜂蜜进行分类。分类方法在来自不同花卉来源的各种蜂蜜样品上进行了试验。对于百里香、大枣、香菜、小檗、金合欢和苜蓿蜂蜜样品，试验了主成分分析结合判别分析（PCA-DA）和偏最小二乘结合判别分析（PLS-DA）。结果表明，蜂蜜的植物来源会影响类黄酮和酚类化合物的分布。例如，枣蜜样品中橙皮素和白杨素的含量最高，而百里香蜂蜜的咖啡酸含量最高；芫荽蜂蜜中槲皮素和对香豆酸的含量最高。为了减少建模的自变量数量，使用了主成分分析 (PCA) 算法。从 PCA 中提取的三个分数有 83.17% 的方差。分类结果表明，PLS-DA成功地用于预测蜂蜜样品的类成员（100%），但PCA-DA给出的正确分类率最低（97%）。