Department of Food Engineering, Faculty of Engineering, Hacettepe University, Beytepe, 06800 Ankara, Turkey.
Food Chem. 2013 Dec 15;141(4):4397-403. doi: 10.1016/j.foodchem.2013.06.061. Epub 2013 Jun 24.
In this study, adulteration of butter with margarine was analysed using Raman spectroscopy combined with chemometric methods (principal component analysis (PCA), principal component regression (PCR), partial least squares (PLS)) and artificial neural networks (ANNs). Different butter and margarine samples were mixed at various concentrations ranging from 0% to 100% w/w. PCA analysis was applied for the classification of butters, margarines and mixtures. PCR, PLS and ANN were used for the detection of adulteration ratios of butter. Models were created using a calibration data set and developed models were evaluated using a validation data set. The coefficient of determination (R(2)) values between actual and predicted values obtained for PCR, PLS and ANN for the validation data set were 0.968, 0.987 and 0.978, respectively. In conclusion, a combination of Raman spectroscopy with chemometrics and ANN methods can be applied for testing butter adulteration.
在这项研究中，使用拉曼光谱结合化学计量学方法（主成分分析（PCA）、主成分回归（PCR）、偏最小二乘（PLS）和人工神经网络（ANNs））分析了黄油与人造黄油的掺假情况。不同的黄油和人造黄油样品以 0%至 100%w/w 的不同浓度混合。PCA 分析用于黄油、人造黄油和混合物的分类。PCR、PLS 和 ANN 用于检测黄油的掺假比例。使用校准数据集创建模型，并使用验证数据集评估开发的模型。PCR、PLS 和 ANN 对验证数据集的实际值和预测值之间的确定系数（R(2)）值分别为 0.968、0.987 和 0.978。总之，拉曼光谱结合化学计量学和 ANN 方法可用于测试黄油掺假情况。
