傅里叶变换近红外光谱结合RVM与新聚类算法鉴别灵芝孢子油多样掺假类别

发布时间：2018-03-02 23:17

本文选题：食品掺假　切入点：FT-NIR　出处：《光谱学与光谱分析》2017年04期 　论文类型：期刊论文

【摘要】：食品掺假种类众多,手段隐蔽,成为食品安全检测一个重要难题。为摆脱传统模型识别食品中是否存在新掺假类别的局限性,实验以纯净的灵芝孢子油和掺杂不同比例花生油、玉米油、薏仁油、地沟油的五种类别为研究对象,采用傅里叶变换近红外光谱(Fourier transform near infrared spectroscopy,FT-NIR)收集12 400~4 000cm~(-1)范围内的近红外光谱。假设掺杂地沟油为新掺假类别,利用前四种类别的校正集样本构建相关向量机(RVM)多分类器,分别对建模的预测集样本和掺杂地沟油样本进行判别,并借助新聚类算法对判别为纯净的灵芝孢子油的样本做进一步分析验证。研究表明,RVM分类器对于建模的预测集样本判别准确率高达93.75%,说明模型有较强的判别能力,但由于模型局限性,掺杂地沟油样品被误判为纯净的灵芝孢子油;在新聚类算法的决策图上,纯净灵芝孢子油校正集和预测集混合样本的聚类中心数为1,而纯净灵芝孢子油校正集和掺杂了地沟油混合样本聚类中心数为2,直观验证判别结果的准确性。结果表明利用FT-NIR技术结合RVM分类器与新聚类算法对于灵芝孢子油掺假能够有效识别,并且能够定性识别新型掺假类型,为解决食品掺假多样化问题提供一种新思路。
[Abstract]:Food adulteration has become an important problem in food safety detection. In order to get rid of the limitation of traditional model to identify whether there is a new adulteration category in food, pure Ganoderma lucidum spores oil and peanut oil with different proportion were used in the experiment. Five kinds of corn oil, barley oil and gutter oil were studied. The near infrared spectra were collected in the range of 12 400 ~ 4 000 cm ~ (-1) by Fourier near infrared (FT-NIRR). The adulterated gutter oil was assumed to be a new adulterated category. The correlation vector machine (RVM) multi-classifier was constructed by using the first four kinds of calibration set samples, and the prediction set samples and the adulterated gutter oil samples were used to distinguish the predicted samples and the adulterated gutter oil samples, respectively. The new clustering algorithm is used to further analyze and verify the sample of Ganoderma lucidum spore oil which is pure. The research shows that the accuracy of the RVM classifier for the prediction set of the model is as high as 93.75, which shows that the model has a strong discriminant ability. However, due to the limitation of the model, the adulterated gutter oil sample was misjudged as pure Ganoderma lucidum spore oil. The cluster center number of mixed samples of pure Ganoderma lucidum spore oil correction set and prediction set is 1, while that of pure Ganoderma lucidum spore oil correction set and mixed samples doped with gutter oil is 2. The results show that FT-NIR combined with RVM classifier and new clustering algorithm can effectively identify the spores oil adulteration of Ganoderma lucidum. It can identify the new type of adulteration qualitatively and provide a new way to solve the problem of food adulteration.
【作者单位】：福州大学电气工程与自动化学院;福建省医疗器械和医药技术重点实验室;福州大学生物科学与工程学院;福建医科大学医学技术与工程学院;
【基金】：福建省教育厅科技项目(JK2014001) 福建省科技厅国际合作项目(2015I003)资助
【分类号】：TS201.6;O657.33

【相似文献】