基于近红外光谱技术的茶籽油掺伪定性鉴别和定量分析

Qualitative Identification and Quantitative Analysis of Adulterated inCamellia Seed Oil based on Near Infrared Spectroscopy Technology

茶籽油是我国特有的高级食用油。近年来茶籽油掺伪现象层出不穷,研究利用傅里叶变换近红外(FT-NIR)光谱与化学计量学相结合,获得一种快速对不同掺假类型(低芥酸菜籽油、大豆油、二元混合油)和不同掺假度(0%~100%)的茶籽油进行定性和定量检测的方法。基于近红外差异光谱进行判别分析(DA),DA成功识别了不同掺假度的二元与三元混合茶籽油。通过对手动与自动筛选的不同波段结合不同预处理方法建立偏最小二乘法(PLS)定量分析模型,最佳模型对掺假水平具有良好的预测性能,决定系数(R 2)均大于0.91。校正均方根误差(RMSEC)和预测均方根误差(RMSEP)均接近于0。通过交叉验证,最佳模型的交叉验证相关系数均大于0.98,交叉验证均方根误差均小于0.05,表明筛选的最佳模型均具有良好的稳定性。通过外部验证,最佳模型对不同掺假类型的中、高掺假度(≥10%)的样品识别率高达100%。

Camellia seed oil is a unique high-grade edible oil in China.In recent years,camellia seed oil adulteration has emerged one after another.In this study,Fourier near infrared(FT-NIR)spectroscopy and chemometrics were used to obtain a rapid method for qualitative and quantitative detection of camellia seed oil with different adulteration types(low erucic acid rapeseed oil,soybean oil,binary mixed oil)and different adulteration degrees(0%~100%).Linear discriminant analysis(LDA)was carried out based on near infrared difference spectrum.LDA successfully identified binary and ternary mixed camellia seed oil with different adulteration degrees.The partial least squares(PLS)quantitative analysis model was established by combining different preprocessing methods with different bands screened manually and automatically,the best model had good prediction performance for adulteration level,and the determination coefficient(R 2)was greater than 0.91.Both corrected root mean square error(RMSEC)and predicted root mean square error(RMSEP)were close to 0.Through cross validation,the cross-validation correlation coefficients of the best models were greater than 0.98,and the root mean square error of cross validation was less than 0.05,indicating that the selected best models had good stability.Through external verification,the recognition rate of the best model for samples with medium and high adulteration degree(≥10%)of different adulteration types was as high as 100%.