为充分提取复杂掺伪食用油的特征信息, 提出并建立一种掺伪芝麻油的判别方法.采集40个纯芝麻油和40个掺入不同浓度玉米油的芝麻油的常规一维近红外透射光谱和中红外衰减全反射光谱.对两样本采用二维相关谱技术进行相关计算, 得到每一样品的同步和异步二维近红外相关谱和中红外相关谱, 并进行预处理, 得到其对应的同步-异步二维近红外相关谱和中红外相关谱.采用多维主成分分析法提取其特征, 并将其得分矩阵进行融合.基于融合的得分矩阵, 以及单一近红外、中红外相关谱得分矩阵分别建立纯芝麻油和掺伪芝麻油偏最小二乘判别分析模型, 三个模型对预测集样品的判别正确率分别为100%、96.2%和96.2%.研究结果表明, 所提出的方法可提取更多的特征信息, 提供更好的分析结果.

To extract the characteristic information of complex adulterated edible oils, a method was established to distinguish adulterated sesame oil. The conventional one-dimensional near-infrared transmission spectra and mid-infrared attenuated total reflectance spectra of 40 pure and adulterated sesame oil samples were collected. The technology of two-trace two-dimensional correlation spectroscopy was used to obtain the synchronous and asynchronous two-dimensional near-infrared and two-dimensional mid-infrared correlation spectra of each sample. The corresponding synchronous-asynchronous two-dimensional near-infrared and two-dimensional mid-infrared correlation spectra were obtained by pretreatment. The multi-way principal component analysis was applied to extract feature information, and the score matrices were fused. The partial least squares discrimination analysis models of pure and adulterated sesame oil were established respectively using the fusion score matrix, the score matrix of synchronous-asynchronous near-infrared correlation spectra, and the score matrix of synchronous-asynchronous mid-infrared correlation spectra. The discriminant accuracies of the three models are 100%, 96.2% and 96.2% respectively. The results show that the proposed method can extract more feature information and provide better analysis results.