蜂蜜真伪的近红外光谱鉴别研究

如何有效鉴别蜂蜜真伪是目前我国蜂产品质量控制的难题之一。提出了一种用傅里叶变换近红外光谱结合判别偏最小二乘法(DPLS)快速鉴别蜂蜜真伪的新方法。首先采集了71个商品蜜样品的近红外光谱数据,其中包括27个纯蜂蜜和44个掺假蜂蜜,然后5次随机划分建模集样本和验证集样本,并对建模样本进行不同光谱预处理,选择并优化不同波段范围和主成分数,用DPLS法建立了5组蜂蜜样本的真伪鉴别模型,外部交叉验证法预测模型。通过对结果进行比较分析,5组校正模型中真蜂蜜和掺假蜂蜜的总体识别准确率分别为91.49%,94.68%,92.98%,93.86%,94.87%;预测样本的识别准确率为86.96%～93.75%,其中模型2,3,4中掺假蜜的识别准确率达100%。研究结果表明,该方法可作为鉴别商品蜜真伪的一种快速筛选技术,在我国蜂蜜质量监控中具有重要意义。

Determination of Adulteration in Honey Using Near-Infrared Spectroscopy

The objective of the present research is to study the potential of using Fourier transform near-infrared spectroscopy(FT-NIR) in conjunction with discriminant partial least squares(DPLS) chemometric techniques for the discrimination of honey authenticity.First,seventy one commercial honey samples from Chinese market were analyzed to detect the levels of honey adulteration by stable carbon isotope ratio and the chemical result showed that the samples include unadulterated(n=27) and adulterated(n=44) products.The samples were scanned in the spectral region between 4 000 and 11 000 cm-1 by FT-NIR spectrometer with an optic fiber of 2 mm path-length and an InGaAs detector and then divided randomly five times into two sets,namely calibration sets and validation sets,respectively.Five kinds of mathematic models of honey samples were established for classification of honeys as authentic or adulterated by using DPLS.Different spectra pretreatment methods,spectral range and different principal component factors were selected to optimize the calibration models.The calibration models were successfully validated with exterior cross-validation methods.Through comparison analysis of the results,the overall corrected identification rate of authentic and adulterated honey samples in five calibration models were 91.49%,94.68%,92.98%,93.86% and 94.87%,respectively.The correct classification rate of the validation samples was 93.75%,89.58%,89.29%,92.31% and 86.96% from model one to model five,respectively and 100% of adulterated honey samples were correctly identified and classified in validation models 2,3 and 4.The results demonstrated that FT-NIR together with DPLS could be used as a rapid and cost-efficient screening tool for discrimination of commercial honey adulteration,and the analytical technique would be significant to Chinese honey quality supervision.