论文部分内容阅读
对炮制前后续断散指纹图谱进行模式识别研究,并根据不同炮制方法进行分类判别。建立不同炮制品续断散HPLC指纹图谱共有模式,进行特征选择,然后采用判别分析法(LDA)、簇类独立软模式分类法(SIMCA)、偏最小二乘判别分析法(PLS-DA)分别建立分类模型,并进行交叉验证。结果表明3种方法均取得较高判别率,能有效地对续断散指纹图谱进行分类,其中判别分析法最优。本试验建立的方法实现了续断散生品及不同炮制品的区分,为中药炮制的质量控制提供了有效的参考依据。
The pattern recognition of fingerprints before and after concocted was conducted and the classification was made according to different processing methods. The common mode of HPLC fingerprinting was established for different processed products, and the features were selected. Then, the discriminant analysis (LDA), cluster independent soft pattern classification (SIMCA) and partial least squares discriminant analysis (PLS-DA) Establish a classification model, and cross-validation. The results show that the three methods have achieved a high discrimination rate, which can effectively classify the fingerprints of Broken Brokeback, of which discriminant analysis is the best. The method established in this experiment realizes the distinction between broken loose bulk products and different processed products and provides an effective reference for the quality control of processed Chinese medicines.