论文部分内容阅读
目的:通过超高效液相色谱(UPLC)指纹图谱与化学计量学相结合对生首乌及制首乌进行化学模式识别研究,并探讨炮制前后化学成分变化。方法:建立UPLC指纹图谱,获取峰面积、保留时间,基于主成分分析(PCA)建立无监督模式识别模型,根据其结果建立4种有监督模式识别模型,并采用对偶传播人工神经网络(CP-ANN)获取的权重值分析炮制前后化学成分变化。结果:偏最小二乘法-判别分析(PLS-DA)、支持向量机-判别分析(SVMDA)、对偶传播CP-ANN模型对生品和炮制品的预测成功率均达到了100%,而簇类独立软模式法(SIMCA)预测结果较差;获取了生首乌与制首乌类中的重要化学成分。结论:该方法可对生首乌与制首乌进行识别与预测,并能从整体化学成分角度研究化学成分变化。
OBJECTIVE: To study the chemical pattern recognition of raw Polygonum multiflorum and Radix polygoni multiflori by UPLC fingerprinting combined with chemometrics, and to explore the chemical composition changes before and after processing. Methods: UPLC fingerprinting was established to obtain the peak area and retention time. An unsupervised pattern recognition model was established based on principal component analysis (PCA). Based on the results, four models with supervised pattern recognition were established. The dual-propagation artificial neural network (CP- ANN) to obtain the weight value analysis of chemical composition changes before and after processing. Results: The predicted success rates of PLS-DA, SVMDA and dual Propagation CP-ANN models for both raw and processed products were all 100% The independent soft mode method (SIMCA) predicted poor results; access to the raw Radix and Radix species important chemical composition. Conclusion: This method can identify and predict Radix Polygoni Multiflori and Radix Polygoni Multiflori and study chemical composition changes from the perspective of overall chemical composition.