基于红外光谱曲线的中药材品种辨识方法研究

Research on Identification Method of Traditional Chinese Medicine Varieties Based on Infrared Spectrum Curve

为了探索应用红外光谱技术辨识中药材的品种,以中红外光谱数据为依据,首先,通过离群点分析剔除了奇异样品,使用移动平均法对原始数据做平滑处理,使用标准差法降维,使用差异度评价聚类效果。其次,分别考察了系统聚类法、K-均值聚类法和模糊K-均值聚类法的聚类效果,发现在分为5类的情况下K-均值聚类法的聚类效果最好。第三,从描述分析、箱图、频率分布的角度对5种中药材的特征和差异进行了比较。本方法原理简单,计算高效,结果可靠,对丰富中药材的品种辨识方法具有重要参考价值。

In order to explore the application of infrared spectroscopy technology to identify the varieties of traditional Chinese medicine, the mid-infrared spectroscopy data was used as the basis.Firstly, in order to explore the application of infrared spectroscopy to identify the varieties of traditional Chinese medicine, based on the intermediate infrared spectroscopy data, the strange samples are eliminated through outlier analysis, the moving average method is used to smooth the original data, the standard deviation method is used to reduce the dimension, and the standard deviation difference is used to evaluate the clustering effect. Secondly, the clustering effects of hierarchical clustering method, k-means clustering method and fuzzy k-means clustering method are investigated respectively. Thus, it is found that the clustering effect of k-means clustering method is the best when it is divided into five categories. Thirdly, the characteristics and differences of five kinds of traditional Chinese medicine are compared from the perspective of description analysis, box diagram and frequency distribution. This method has such advantages as simple principle, efficient calculation and reliable results, which is of important valuable reference for enriching the variety identification methods of traditional Chinese medicine.