摘要
目的探讨连续投影算法(SPA)结合偏最小二乘法(PLS)用于快速测定秦皮提取液中秦皮甲素的可行性。方法以HPLC分析值作为参照,采集不同产地的秦皮提取液的紫外光谱,运用SPA算法筛选样本的紫外特征波长,PLS算法建立快速测定秦皮甲素的数学模型。结果应用SPA算法筛选,只用了全部变量的16.21%(65个变量)建立模型,所建模型的预测集相关系数(RP)及预测均方根(RMSEP)分别为0.969 0、0.067 0。结论 SPA算法可以在不降低模型性能的前提下显著降低模型的复杂度,结合PLS算法建立的模型可以用于快速测定秦皮提取液中秦皮甲素。
AIM：To use successive projections algorithm（SPA） and partial least squares（PLS） to fast determine aesculin of Fraxini Cortex extract from different batches.METHODS Ultraviolet spectroscopy of Fraxini Cortex extract was selected.The characteristic spectral data was deduced using SPA and subjected to PLS model to fast determine aesculin with the reference of HPLC.RESULTS Sixty-five（16.21%） characteristic wavelengths were identified by SPA as significant wavelengths used for determination of aesculin content.With the best calibration model built on parameters,the correlation coefficient of prediction data was 0.969 0,and the mean square error of prediction data was 0.067 0.CONCLUSION The model deduced from SPA proves to be good in practice for fast determination of aesculin content of Fraxini Cortex extract.
出处
《中成药》
CAS
CSCD
北大核心
2011年第5期847-851,共5页
Chinese Traditional Patent Medicine
基金
上海中医药大学自然科学基金资助项目(0934)
上海市教委高校选拔培养优秀青年教师科研专项基金(5ZY10100)
关键词
连续投影算法
偏最小二乘法
紫外光谱
秦皮甲素
successive projections algorithm
partial least squares
ultraviolet spectra
aesculin
Fraxini Cortex
