基于PCA-SVM结合共聚焦拉曼光谱的特级初榨橄榄油掺伪压榨菜籽油定量分析

Quantitative analysis of extra virgin olive oil adulterated with pressed rapeseed oil based on PCA-SVM and confocal Raman spectroscopy

为了促进国内橄榄油市场的健康发展,对掺伪同样存在天然类胡萝卜素的低温压榨菜籽油的特级初榨橄榄油进行了定量鉴别研究。采用共聚焦拉曼光谱技术对不同掺伪浓度油样进行测试,基于密度泛函理论对油样的拉曼光谱峰的归属进行了理论分析,并对拉曼光谱数据进行主成分分析(PCA),然后利用支持向量机(SVM)构建PCA-SVM模型。另外,对PCA-SVM模型的检出限进行了研究。结果表明:特级初榨橄榄油与低温压榨菜籽油的拉曼光谱存在一定差异,最明显的光谱差异主要集中在谱峰1008、1161、1528 cm^(-1)和谱段2800~3000 cm^(-1)内,与密度泛函理论对不同油样拉曼光谱峰的分析一致;不考虑类胡萝卜素特征信号建立的PCA-SVM模型决定系数大于0.989,均方根误差小于2.990%,检出限为2%(低温压榨菜籽油体积分数);在特级初榨橄榄油掺伪定量分析中,考虑类胡萝卜素的特征信号有助于提高模型预测精度,但仅限于掺伪低价植物油中无类胡萝卜素存在的情况;PCA-SVM模型在不考虑类胡萝卜素特征信号的情况下依然具有良好的定量预测效果。综上,所建立的PCA-SVM模型可以用于掺伪2%以上低温压榨菜籽油的特级初榨橄榄油的定量鉴别。

To promote the healthy development of the domestic olive oil market,a study was conducted on quantitatively identifying extra virgin olive oil adulterated with low-temperature pressed rapeseed oil containing natural carotenoids.The confocal Raman spectroscopy technology was used to test oil samples with different adulteration amounts.Theoretical analysis was conducted on attributing Raman spectral peaks of different oil samples based on density functional theory.Principal component analysis(PCA)was performed on the Raman spectral data,and then the support vector machine(SVM)was used to construct a PCA-SVM model.In addition,the detection limit of the PCA-SVM model was studied.The results showed that there was difference in the Raman spectra of extra virgin olive oil and low-temperature pressed rapeseed oil,and the most apparent spectral differences mainly concentrated in the spectral peaks 1008,1161,1528 cm^(-1) and spectral bands 2800-3000 cm^(-1),which was consistent with the analysis of Raman spectral peaks of different oil samples using density functional theory.The PCA-SVM model established without considering the characteristic signals of carotenoids had a coefficient of determination greater than 0.989,and the root mean square error was lower than 2.990%,and the detection limit was 2%(volume fraction of low-temperature pressed rapeseed oil).In the quantitative analysis of extra virgin olive oil adulteration,considering the characteristic signals of carotenoids could help to improve the prediction accuracy of the model,but it was only limited to the absence of carotenoids in adulterated low-price vegetable oils.The PCA-SVM model still had good quantitative prediction performance even when the characteristic signals of carotenoids were not considered.In summary,the established PCA-SVM model can be used to quantitatively identify extra virgin olive oil adulterated with over 2%low-temperature pressed rapeseed oil.