应用可见/近红外光谱进行黄酒品种的判别

为了实现对黄酒品种的快速判别，采用可见/近红外光谱对不同品种的黄酒获取光谱曲线，然后采用主成分分析方法对光谱数据进行聚类分析，并将其提取的主成分作为BP神经网络的输入值，建立了黄酒品种鉴别模型。该模型将前6个主成分作为神经网络的输入变量，加速了神经网络的学习速度，提高了模型的预测精度。随机选取每个品种的15个黄酒样本，共45个样本组成预测集，剩余的145个黄酒样本组成训练集建立训练模型，并用预测集样本对其进行验证。将品种鉴别的偏差标准定为±0.1，结果表明，只有1个未知样本超出偏差范围，该方法的品种鉴别正确率为97.78％，获得了满意的结果。说明文章提出的方法具有很好的分类和鉴别作用，为黄酒品种的快速鉴别提供了一种新方法。

Discrimination of Varieties of Yellow Wines Using Vis/NIR Spectroscopy

In order to achieve the rapid discrimination of the varieties of yellow wines, the spectral curves of yellow wines were obtained by Vis/NIR spectroscopy, and the principal component analysis (PCA) was applied to perform the clustering analysis. The principal components (PCs) extracted by PCA were employed as the inputs of the BP neural networks, and then a discrimination model was built. The first 6 PCs were regarded as the new eigenvectors to accelerate the training speed and to improve the precision of the model. Fifteen samples from each variety and a total of 45 samples were selected randomly as the prediction sets. The remaining 145 samples were used as the training sets to build the training model which is validated by the samples of the prediction sets. The result error was set to be +/- 0.1, and the results indicated that only one sample exceeded the threshold value, therefore the recognition rate of 97.78% and an excellent precision were achieved. So the discrimination method studied in the present paper played a good role in the classification and discrimination, and offered a new approach to the rapid discrimination of the varieties of yellow wines.