摘要
通过Soft IndependentModeling ofC lassAnalogy(SIMCA)模式识别方法区分花生油、大豆油、米糠油、棕榈油、菜籽油、玉米油、棉籽油、葵花籽油和芝麻油9种植物油脂.采用气相色谱法分析9种植物油脂219个样品的脂肪酸,用面积归一化法得到每个植物油脂的各脂肪酸相对含量.以每种植物油脂中9个脂肪酸的相对含量为变量,采用SIMCA分析技术进行数据预处理,随机取2/3的样品作定标集,1/3作验证集,对9种植物油脂的训练集进行主成分分析(PCA),并通过交互验证建立各油脂种类的PCA模型,再利用训练集样本建立的SIMCA判别模型对验证集样本进行验证.结果显示,SIMCA可以对9种植物油脂分别聚类和识别,各种植物油脂的SIMCA分析的聚类精度均为100%,除了芝麻油的验证识别准确率为75%外,其他均为100%.
In this work, soft independent modeling of class analogy （SIMCA）pattern recognition analysis was applied to discriminate peanut oil, soybean oil, rice bran oil, palm oil, rapeseed oil, corn oil, cottonseed oil, sunflower oil and sesame oil. The fatty acid composition of 219 samples from 9 kinds of vegetable oil and fat were analyzed by Gas chromatography, and the content of fatty acids was obtained by the peak area normalization method. Each type of vegetable oil in 9 of the relative content of fatty acids as the variable was used in the processing of spectra pretreated, which 2/3 of samples were selected for calibration, the other for validation. Nine kinds of vegetable oil on the training set of principal component analysis （PCA） and through the establishment of cross - validation of the PCA model of the types of vegetable oils and fats, then use SIMCA dis- criminant model established set of samples to verify authentication. The results show that the classification accuracy of the 9 kinds of vegetable oil and fat were yielded about 100%. In addition to validation accuracy of sesame oil were yielded about 75%. The validation accuracy of the 8 kinds of vegetable oil and fat were yielded about 100%.
出处
《河南工业大学学报（自然科学版）》
CAS
北大核心
2009年第5期13-17,共5页
Journal of Henan University of Technology：Natural Science Edition
基金
河南工业大学校科研基金项目(07XJC001)
关键词
植物油脂
脂肪酸
气相色谱
判别
SIMCA
PCA
vegetable oil and fat
fatty acid
gas chromatography
discriminant
SIMCA
PCA
