基于FTIR的芝麻油真伪鉴别和掺伪定量分析模型

把低价油掺入到高价油是食用油脂中的常见掺伪现象，芝麻油由于品质好价格高，市场上时有假冒伪劣产品，因此应用FTIR并结合化学计量学，建立了芝麻油的真伪和掺伪的快速分析方法。首先分析了芝麻油与大豆油、葵花籽油在4 000～650 cm-1范围的FTIR谱图，由于食用植物油都是不同脂肪酸甘油三酯的混合物，其谱图极为相似，很难发现芝麻油与其他油脂的明显差异。但是不同食用油的脂肪酸组成不同，其1 800～650 cm-1红外指纹特征区也有所不同，因此可以选择该区域，对红外光谱数据用化学计量学方法进行分类识别。通过建立主成分分析(PCA)和簇类独立软模式识别(SIMCA)模型，进行了芝麻油的真伪鉴别，该模型聚类效果较为理想，识别正确率达到了100％;采用标准正态化校正(SNV)和偏最小二乘法(PLS)，经过PCA分析计算，芝麻油中掺入大豆油、葵花籽油的掺伪检测限均为10%;利用FTIR和PLS，建立了芝麻油掺的定量分析模型，该模型预测值与实际值有着良好的对应关系，预测相对误差为-6.87％～8.07％之间，说明定量模型可行。本方法能够实现芝麻油的快速真伪鉴别和掺伪定量分析，其优点是模型一旦建立，分析简便、快速，可以满足大量样品的日常监测。

Authentication and Adulteration Analysis of Sesame Oil by FTIR Spectroscopy

It’s common in edible oil market that adulterating low price oils in high price oils. Sesame oil was often adulterated because of its high quality and price, so the authentication and adulteration of sesame oil were qualitatively and quantitatively analyzed by Fourier transform infrared (FTIR) spectroscopy combined with chemometrics. Firstly, FTIR spectra of sesame oil, soybean oil, and sunflower seed oil in 4 000～650 cm-1 were analyzed. It was very difficult to detect the difference among the spectra of above edible oils, because they are all mixtures of triglyceride fatty acids and have similar spectra. However, the FTIR data of edible oils in the fingerprint region of 1 800～650 cm-1 differed slightly because their fatty acid compositions are different, so the data could be classified and recognized by chemometric methods. The authenticity model of sesame oil was built by principal component analysis (PCA) and soft independent modeling of class analogy (SIMCA). The recognition rate was 100%, and the built model was satisfactory. The classification limits of both soybean oil and sunflower seed oil adulterated in sesame oil were 10%, with the chemometric treatments of standard normal variation (SNV), partial least square (PLS) and PCA. In addition, the FTIR data processed by PCA and PLS were used to establish an analysis model of binary system of sesame oil mixed with soybean oil or sunflower oil, the prediction values had good corresponding relationship with true values, and the relative errors of prediction were between -6.87% and 8.07%, which means the quantitative model was practical. This method is very convenient and rapid after the models have been built, and can be used for rapid detection of authenticity and adulteration of sesame oil. The method is also practical and suitable for the daily analysis of large amount of samples.