【摘要】 取48批次水冬瓜根皮鲜药材于50℃分别烘干后粉碎至通过孔径0.23mm的样筛。混匀后装于样品瓶中制备成48个样品,编号为G1～G48,其中42个样品(G1～G42)作为校正集,6个样品(G43～G48)为验证集。应用声光可调滤光器(AOTF)-近红外光谱仪(NIR)采集每个样品的原始光谱。确定NIR光谱预处理方式为一阶微分导数,在1 100～2 200nm波段构建该药材中总黄酮及浸出物含量快速测定的分析模型。采用化学计量学分析软件将校准集样品的光谱信息和上述两测定项目的参考值作数据关联,分别建立该药材中总黄酮和浸出物快速测定的分析模型。用G1样品重复采集其NIR光谱6次,按上述方法作预处理及测定,结果获得总黄酮及浸出物测定值的相对标准偏差(n=6)分别为3.9%,2.8%。仍取G1样品于12h内先后采集其红外光谱8次,按上述相同方法分析,两项测定结果的相对标准偏差分别为2.1%,2.3%,说明12h内样品的稳定性良好。最后将6个验证样品的原始光谱经预处理所得的数据导入化学计量学分析软件作模型预测,两项目的预测值与参考值的相对误差分别在1.6%～6.3%和0.60%～4.3%之间。表明所建模型验证效果良好,应用此方法分析单个样品只需2～5s。更多还原

【Abstract】 Samples of the Miao medicine,Toricellia Angulata Oliv.var.intermedia(Harms)Hu,were collected from 48 batches of the freshcrude drug,and the drugs were pretreated by drying at 50 ℃,powdering to pass through sieve with bore of 0.23 mm and mixing thoroughly to give 48 samples of the drug,numbered from G1 to G48,among which 42 samples(G1 to G42)were taken as calibration set while the remainder 6 samples(G43 to G48)were used as testing set.Initial NIR spectra of the 48 samples were taken by the AOTF-NIR.The intial spectra of samples of the calibration set were pretreated by the 1 st order differentiation in the wavelength range of1 100 to 2 200 nm.The NIR informations of these samples were correlated with the referential data of total flavonoids(TFs)and extract(EXs)in the samples by using the chemometrical soflware,and analytical models for rapid determination of TFs and EXs in the Miao drug were established.Six NIR spectra of a same sample(G1)were taken and treated in the same way as described above to test for precision of the method.Values of RSDs(n=6)found for TFs and EXs were 3.9% and 2.8% respectively.Seven NIR spectra of the sample G1 were taken within the interval of 12 h,and treated in the same way to test for stability of the sample,giving RSDs(n=7)of 2.1%(for TFs)and 2.3%(for EXs).Initial NIR spectra of the 6 samples(G43 to G48)of the testing set were pretreated and predicted by the analytical models,giving values of relation deviations between predicted values and referential values in the range of 1.6%to 6.3%for TFs and 0.60%to 4.3%for EXs,showing good verification results of the established models.It was proved that it takes only 2 to 5 sfor analysis of a single sample.更多还原