摘要
采用拉曼光谱技术对食用油的种类建立定性分析模型,实现快速、无损地识别食用油的种类。选取3种常用食用油(大豆油、花生油、玉米油)共87个样品采集其拉曼光谱,采用一阶导数的方法对光谱进行预处理,Norris导数法进行滤波去噪,处理后的光谱采用判别分析算法建立食用油种类识别模型,模型能够实现3种食用油的准确分类。选取大豆油、花生油、玉米油各5个样品作为测试样品,测试结果为大豆油、花生油和玉米油都能够正确地分入所属类别。结果表明,拉曼光谱结合判别分析的方法能够实现食用油种类的识别,校正模型的分类结果能达到100%,预测样本分类结果准确率为86.7%。
Raman spectroscopy is applied for rapid and non-destructive identification and classification of edible oils including soybean oil, peanut oil and corn oil （n=40）. After using first derivative for preprocessing and Norris derivative for filtering the noise, we set up a qualitative analysis model with the discriminant analysis algorithm to identify the category of edible oils. The model can classify the edible oil correctly. And the test samples, every category of edible oils （soybean oil, peanut oil, corn oil） include 5 samples, can be classified into the right categories. As a result, Raman spectroscopy with discriminant analysis algorithm can be applied to identify the category of edible oils. The accuracy of the calibration model is 100% and the accuracy of the result for classification with prediction samples can reach to 86.7%.
出处
《食品科技》
CAS
北大核心
2015年第3期274-278,共5页
Food Science and Technology
基金
北京市教委科技发展重点项目(KZ201310011012)
北京市教委科技创新平台项目(PXM_2012_014213_000023)
北京市自然科学基金项目(4132008)
关键词
拉曼
食用油
判别分析
定性识别
Raman
edible oil
discriminant analysis
qualitative identification
