基于近红外光谱的掺伪花生油鉴别模型优化

Optimization of Identification Model of Adulterated Peanut Oil Based on Near Infrared Spectroscopy

为实现掺伪花生油的快速鉴别,基于近红外光谱技术,利用偏最小二乘法(PLS法)建立掺伪花生油鉴别模型,并采用不同预处理方法进行模型优化。研究结果表明:利用PLS法建立的模型,对于花生油中掺入大豆油样品的鉴别,在漫反射方式下、采用二阶导数谱、Norris平滑方法、附加散射矫正光程方式、因子数为6时最为理想,其预测集相关系数为0.967 9;对于花生油中掺入菜籽油样品的鉴别,在漫反射方式下、采用二阶导数谱、S-G平滑方法、光程不矫正、因子数为5时最为理想,其预测集相关系数为0.994 8。该分析模型可以为花生油品质监控和快速定量鉴别掺伪提供参考。

Near infrared spectroscopy(NIR)combined with partial least squares(PLS)was employed to identify adulterated peanut oil.Different pretreatment methods were used to optimize the model.The results indicated that for the identification of peanut oil adulterated with soybean oil,the best model established by PLS was under diffuse reflection mode,using secondorder derivative spectrum,Norris smoothing method and additional scattering correction light path method,and the factor was6,in which the correlation coefficient of prediction set(r_p)was 0.967 9.For the identification of peanut oil adulterated with rapeseed oil,the ideal model was under diffuse reflection condition,second-order derivative spectrum,S-G smoothing method and no correction of optical path,and the factor was 5,in which the r_p was 0.994 8.The model could be used in the rapid analysis adulteration in peanut oil.