山茶油素有“东方橄榄油”美誉, 实现掺假山茶油的鉴别具有重要实用价值, 采用近红外光谱技术对掺有葵花油的山茶油进行检测。 分别以1%, 5%, 10%为梯度制备掺假比例不同的山茶油样品, 并根据掺假比例将其分为A组（0%～5%）和B组（6%～10%）共11个样品, C组（15%～40%）6个和D组（50%～100%）6个样品。 将每个掺假样品充分混匀后再分为9份, 依次采集其1 000～2 500 nm范围的吸收光谱, 共获得207条光谱曲线。 每组样品的光谱数据按2∶1随机分为训练集与验证集。 经去除首尾噪声后, 通过主成分分析法（principal component analysis, PCA）降维, 并利用前四个主成分建立了鉴别山茶油不同掺假等级的主成分-支持向量机判别模型, 训练集与验证集的总体判别准确率分别达96.38%和94.20%; 进一步, 通过对前四个主成分的载荷系数的分析, 并结合原始光谱, 提取建模过程中权重较大的波长并解析其化学含义, 最终确定出五个特征波长: 1 212, 1 705, 1 826, 1 905及2 148 nm, 以此波长重新建立近红外特征光谱山茶油掺假等级判别模型, 对训练集与验证集的总体判别准确率也达到了94.20%和92.75%。 研究结果表明, 利用近红外光谱和特征光谱均能够较好实现山茶油掺假等级的鉴别, 同时所建立的近红外特征光谱模型也为设计相应的掺假山茶油实用便携式检测仪器提供了理论基础。

Camellia oleifera oil has the reputation of “oriental olive oil”; it is important to detect the adulterated camellia oleifera oil. In this paper, NIR spectra were used to detect camellia oleifera oil adulterated with sunflower oil. Camellia oleifera oil adulterated with varying mass fraction of sunflower oil were prepared, i. e., 11 samples in 0%～10% with the gradient of 1%, 6 samples in 15%～40% with the gradient of 5%, 6 samples in 50%～100% with the gradient of 10%, and all the samples were divided into four groups such as A(0%～5%), B(6%～10%), C(15%～40%) and D(50%～100%). A total of 207 absorbance spectra(1 000～2 500 nm) were acquired by sampling 9 times in each adulteration. Calibration set was consist of two-thirds of the spectra data in each group selected randomly, and the validation set was made up of the last spectral data. After removing the noise in both ends of the spectra, principal component analysis(PCA) was used to reduce the dimensionality, then the first four PCs were used to build the support vector machine (SVM) identification model, and the identification accuracies of 96.38% and 94.20% in calibration and validation set were obtained. Furthermore, five characteristic wavelengths (1 212, 1 705, 1 826, 1 905 and 2 148 nm) were selected based on the loading of the PCs, the peaks or troughs of the original spectra and the chemical functional groups they were corresponding to. A NIR simplified SVM identification model was built by them, and the identification accuracies were 94.20% and 92.75%. Overall, both NIR spectroscopy and NIR characteristic spectra can realize the identification of camellia oleifera oil adulterated with sunflower oil, and the characteristic wavelengths, selected in this study, provide a basis for the design of corresponding instrument.