对掺入不同含量大豆油和菜籽油的鱼油进行鱼油掺假含量的可见-近红外光谱(Vis-NIR)研究。 向3个不同品牌鱼油中分别掺入不同比例的大豆油, 另外3个不同品牌中分别掺入不同比例的菜籽油, 共获得300个样本。 对所采集样本的光谱数据分别采用原始光谱, 以及平滑, 变量标准化(SNV), 多元散射校正(MSC), 一阶求导和二阶求导等预处理算法进行处理后, 建立偏最小二乘回归(PLSR)模型。 基于全波段光谱的鱼油中大豆油和菜籽油掺假含量预测的最优模型分别为全波段PLSR模型和MSC-PLSR模型, 其预测相关系数(Rp)分别达到0.938 6和0.959 3。 进一步采用连续投影算法(SPA)分析鱼油中大豆油和菜籽油掺假样品的光谱, 并分别获得了11个和15个光谱特征波长变量。 基于特征变量的PLSR模型的Rp分别为0.941 2和0.932 6。 试验研究表明, 可以采用Vis-NIR技术实现对鱼油掺假物含量的检测。

Visible and near infrared (Vis-NIR) spectroscopy was used to accomplish a rapid and noninvasive quantification of the two common adulterants, soybean oil and rapeseed oil, in fish oil. Different contents of soybean oil were added into fish oil of three brands and different contents of rapeseed oil were added into fish oil of another three brands, the Vis-NIR spectra of adulterated samples were collected, pretreated by five spectral preprocessing algorithms (smoothing, standard normal variate (SNV), multiplicative scatter correction (MSC), 1st-derivative, and 2nd- derivative), and used to establish partial least square regression (PLSR) models. The correlation coefficients for prediction (Rp) of 0.938 6 and 0.959 3 were obtained for the adulterant detection of soybean oil and rapeseed oil respectively, and their optimal models were full range spectral PLSR model and MSC-PLSR model. Successive projections algorithm (SPA) was then used to analyze the full range spectra of fish oil samples adulterated with soybean oil and rapeseed oil respectively, and 11 and 15 spectral characteristic wavelength variables were obtained. The Rp of 0.941 2 and 0.932 6 were obtained based on SPA-PLSR models for the adulterant detection of soybean oil and rapeseed oil, respectively. The overall results indicate that Vis-NIR spectroscopy is a feasible way to determine the adulterants of soybean oil and rapeseed oil in fish oil rapidly and non-destructively.