模式识别法分析5种植物油脂

Pattern Recognition Analysis of 5 Kinds of Vegetable Oil and Fat by the Content of Fatty Acids

通过模式识别方法区分花生油、大豆油、米糠油、棕榈油和菜籽油。采用气相色谱法分析5种植物油脂的脂肪酸,用面积归一化法计算每个植物油脂样品的各脂肪酸相对含量。以每个植物油脂中9个脂肪酸的相对含量为变量,采用SPSS13.0软件的模式识别技术对119个植物油脂样品进行区分。由主成分分析图可知,花生油、大豆油、米糠油、棕榈油和菜籽油被清晰地分为5组。判别分析建立的判别方程能较好地实现样品的判别,自身验证和交互验证的准确率均为100%。另取每种植物油脂各5个样品(共25个)进行验证,识别准确率为100%。对调和有棕榈油的花生油进行主成分分析,在主成分分析图上,调和油的分布点在花生油分布区域与棕榈油分布区域之间。

Principal component analysis （PCA） was apphed to discriminate peanut oil, soybean oil, ricebran oil, palm oil and rapeseed oil. The fatty acid composition of 119 samples from 5 species of vegetable oil and fat was determinated by gas chromatography. Discriminant analysis using the pattern recognition technique of SPSS13.0 was able to classify the samples as pure oil and fat based on their fatty acids. It showed that 119 samples were clustered in distinct groups and each group could be clearly separated in the principal component analytical plot. A model for discriminant analysis was developed, and the accuracy of self-validation and crossvalidation was 100%, respectively. The data of the study show that a 100% prediction ability was obtained to the species of 25 unknown samples. It was found that the distribution points of peanut oil adulterated with varying concentrations of palm oil were between the region of peanut oil and palm oil in the principal component analytical plot.