基于Fisher判别法的中药材品种与产地的鉴别方法研究

Study on Identification Methods of Varieties and Origin of Traditional Chinese Herbal Medicine Based on Fisher Discriminant Approach

为了探索应用红外光谱技术鉴别中药材的品种和产地,本文以红外光谱数据为依据,首先通过离群点分析剔除奇异样品,使用移动平均法对原始数据做平滑处理,使用一阶差分或高阶差分消除分化现象,使用标准差法降维。其次使用经典的Fisher判别法对中药材的品种和产地进行了判别,模拟精度达到了100%,预测精度在89.23%~100%之间。

In order to explore the application of infrared spectroscopy to identify the varieties and origin of traditional Chinese medicine, based on the infrared spectrum data, firstly, the strange samples are eliminated through outlier analysis, the original data are smoothed by moving average method, the differentiation phenomenon is eliminated by first-order difference or high-order difference, and the dimension is reduced by standard deviation method. Secondly, the classical Fisher discriminant approach is used to distinguish the varieties and origin of traditional Chinese medicine. The simulation accuracy reaches 100 percent and the prediction accuracy is between 89.23 percent and 100 percent.