应用可见-近红外光谱技术进行白醋品牌和pH值的快速检测

提出了一种基于可见-近红外透射光谱技术快速判别白醋品牌和测定pH值的方法。应用可见-近红外透射光谱获取不同品牌白醋的透射光谱曲线，并对获得的原始光谱数据进行平滑、变量标准化以及一阶导数等预处理，然后利用主成分分析对原始光谱数据进行聚类分析，根据主成分的累计贡献率选取主成分数，并将所选取的主成分作为三层BP神经网络的输入。通过定标集样本对BP神经网络进行训练，得到三层优化神经网络结构：5输入层节点，6隐含层节点和2输出层节点，各层传递函数均采用Sigmoid函数。利用该模型对预测集样本进行预测。实验结果表明在阈值设定为±0.1的情况下该模型对预测集样本品牌鉴别准确率达到了100%，pH预测值与实际测量值偏差小于5%，得到了理想的结果。所以利用可见-近红外光谱技术结合主成分分析和神经网络算法能够快速准确的判定白醋品牌和pH值。

Fast Detection of White Vinegar Varieties and pH by Vis/NIR Spectroscopy

White vinegar is a condiment indispensable in our life, but our understanding of the white vinegar and evaluation of its quality and function has been gained through routine chemical and physical analysis. It is called for to develop more time- and cost-efficient methodologies for white vinegar detection. Visible and near infrared spectroscopy (Vis/NIR) is a nondestructive, fast and accurate technique for the measurement of chemical components based on overtone and combination bands of specific functional groups. Vis/NIR transmittance spectroscopy and chemometrics methods were utilized in classification and pH mensuration of white vinegar in the present study. First, the spectral curves of white vinegar were obtained by handheld Vis/NIR spectroradiometer, then principal component analysis (PCA) was used to process the spectral data after pretreatment. Five principal components (PCs) were selected based on accumulative reliabilities (AR), and these selected PCs would be taken as the inputs of the three-layer back-propagation artificial neural network (BP-ANN). A total of 240 white vinegar samples were divided into calibration set and validation set randomly, the calibration set had 180 samples with 60 samples of each variety, and the validation set had 60 samples with 20 samples of each variety. The BP-ANN was trained using samples in calibration set, the optimal three-layer BP-ANN model with 5 nodes in input layer, 6 nodes in hidden layer, and 2 nodes in output layer would be obtained, and the transfer function of sigmoid was used in each layer. Then, this model was used to predict the samples in the validation set. The result indicated that a 100% recognition ration was achieved with the threshold predictive error +/- 0.1, the bias between predictive value and standard value was lower than 5%. It could be concluded that PCA combined with BP-ANN was an available method for varieties recognition and pH mensuration of white vinegar based on Vis/NIR transmittance spectroscopy.