论文部分内容阅读
目的:以不同厂家不同批号的一清颗粒样品作为分析对象,采用近红外漫反射与判别分析相结合的方法建立聚类模型。方法:对扫描光谱进行多元散射校正(MSC)+一阶导数处理,选择18个主成分数进行计算,建立定性模型。结果:应用判别分析法能够对不同厂家的一清颗粒进行鉴别,并对其进行快速分类,分类结果与实际所属基本一致。结论:该模型可直接对样品进行快速准确的分类,可以应用于不同厂家一清颗粒制剂的分类鉴别。
OBJECTIVE: To establish a clustering model by combining near-infrared diffuse reflectance and discriminant analysis with Yiqing particle samples from different manufacturers with different batch numbers. Methods: Multi-scatter correction (MSC) + first-order derivative processing was performed on the scanning spectrum. 18 principal component numbers were selected for calculation to establish a qualitative model. Results: Discriminant analysis method can be used to identify different manufacturers of a clear particles, and its rapid classification, the classification results are basically the same with the actual belonging. Conclusion: The model can be used to quickly and accurately classify samples, and it can be applied to the classification and identification of different manufacturers’ Yiqing granules.