School of Chinese Material Medica, Beijing University of Chinese Medicine, Beijing 100029, China.
Research Institute of Beijing Tongrentang Co., Ltd., Beijing 100079, China.
Food Chem. 2020 Dec 1;332:127356. doi: 10.1016/j.foodchem.2020.127356. Epub 2020 Jun 21.
This study is about the rice syrup adulteration determination in different botanical origin honey in the food product. Due to time-consuming and large risk of misdiagnosis, it is essential to establish a general model for adulteration detection regardless of the original category of honey. In this paper, infrared (IR) spectra combined with four supervised pattern recognition methods were employed to establish the general model for rice syrup adulteration detection in acacia, linden and jujube honey samples simultaneously. Moreover, Monte-Carlo sampling technology was executed to evaluate the models via the average accuracy, sensitivity and specificity. The first derivative-least squares support vector machines (Der-LS-SVM) gave an outstanding performance with higher accuracy (97.09%), higher sensitivity (96.64%), higher specificity (97.58%) and lower standard deviations after fifty trials. In addition, this study makes further efforts to control the quality of the honey product in the market on rice syrup adulteration.
本研究旨在检测食品中不同植物源蜂蜜中的大米糖浆掺假情况。由于耗时且误诊风险较大，因此建立一种通用的掺假检测模型非常重要，而不论蜂蜜的原始类别如何。在本文中，我们同时采用了红外（IR）光谱结合四种监督模式识别方法，建立了一种通用的模型，用于检测在阿拉伯胶、椴树和枣花蜜样本中的大米糖浆掺假情况。此外，我们还采用了蒙特卡罗抽样技术通过平均准确率、灵敏度和特异性来评估模型。一阶导数最小二乘支持向量机（Der-LS-SVM）的表现非常出色，在五十次试验后，其准确率（97.09%）、灵敏度（96.64%）、特异性（97.58%）更高，标准偏差更低。此外，本研究还进一步努力控制市场上蜂蜜产品中大米糖浆的掺假情况。
