基于近红外光谱的掺伪乳制品鉴别模型优化

Optimization of Identification Model of Adulterated Dairy Products Based on Near Infrared Spectroscopy

为实现掺伪乳制品的快速鉴别,基于近红外光谱技术,采用偏最小二乘法(PLS法)建立掺伪乳制品的鉴别模型,并利用剔除异常点等方法对模型进行优化。结果表明:掺水乳制品在透射方式下,因子数为5,采用原始光谱、S-G平滑、附加散射矫正方式时定量模型最为理想,其预测集相关系数为0.9975,决定系数为0.9943;掺淀粉乳制品在透射方式下,因子数为4,采用一阶导数谱、Norris平滑、不矫正时定量模型最为理想,其预测集相关系数为0.9913,决定系数为0.9827。

Near infrared spectroscopy(NIR)combined with partial least squares(PLS)was used to identify rapidly adulterated dairy products.Eliminating abnormal points and other methods were em-ployed to optimize the model.The results showed that for the identification of dairy products adulterated with water,the ideal model established by PLS was under transmission mode,using Spectrum,S-G smoothing method and additional scattering correction light path method,and the factor was 5,in which the correlation coefficient of prediction set(RP)was 0.9975 and the coeffi-cient of determination(R2)was 0.9943.For the identification of dairy products adulterated with starch,the best model was under transmission condition,first-order derivative spectrum,Norris smoothing method and no correction of optical path,and the factor was 4,in which the RP was 0.9913 and the R2 was 0.9827.