论文部分内容阅读
目的 :为元胡止痛散建立一种快速有效的定量分析方法 ,并为将近红外光谱分析技术应用于中药的定量分析提供指导。方法 :按处方配制 2 5个模拟样本 ,随机挑选 1 8个组成训练集 ,另外 7个组成预示集 ,采集各样本的近红外光谱数据 ,用BP神经网络和PLS法对数据进行处理 ,并实际分析了三批样品。结果 :模拟样本中 ,对于元胡 ,采用BP网络和PLS法 ,平均相对预示误差分别为 1 .5% ,2 .5% ,对于白芷 ,平均相对预示误差分别为 2 .9% ,4 .4 % ,对于实际样本 ,各组分标示量的百分含量都在 95%～ 1 0 5%之间。结论 :近红外光谱结合BP神经网络或PLS应用于元胡止痛散的定量分析是可行和有效的
Objective : To establish a rapid and effective quantitative analysis method for Yuanhu Zhitongsan, and to provide guidance for the application of near infrared spectroscopy in quantitative analysis of traditional Chinese medicine. Methods: According to the prescription, 25 simulation samples were prepared, 18 training sets were randomly selected, and 7 other prediction sets were collected. Near-infrared spectral data of each sample was collected, and data was processed by BP neural network and PLS method. Three batches of samples were analyzed. Results: In the simulation sample, for Yuan Hu, using BP network and PLS method, the average relative prediction errors were 1.5% and 2.5%, respectively, and for the daytime, the average relative prediction error was 2.9%, respectively. % For the actual sample, the percentage content of each component is between 95% and 10%. Conclusion : It is feasible and effective to apply quantitative analysis of near-infrared spectroscopy combined with BP neural network or PLS to Yuanhu analgesic powder.