近红外漫反射光谱法测定不同厂家一清颗粒的含量

Quantitative analysis on Yiqing granule from different manufacturers by near infrared diffuse reflectance spectroscopy

目的:应用近红外漫反射光谱结合偏最小二乘法,建立25个厂家85批一清颗粒中黄芩苷含量的近红外光谱定量分析模型。方法:在12000～4000cm-1光谱范围内扫描样品,以校正均方差(RMSEC)和相关系数(R2)为指标,通过筛选,确定了用于建模的最优近红外波段和光谱预处理方法。采用偏最小二乘法建立了近红外光谱与HPLC分析值之间的校正模型,并以此预测了10个未知样本。结果:近红外预测值与真实值的相关系数R2为0.9732,校正均方差(RMSEC)为0.0625,含量预测回收率为101.1%,RSD为3.0%,预测均方差(RMSEP)为1.5031。结论:该模型可直接对样品进行快速准确的检测,具有非破坏性、无污染、重现性好等优点,可以应用于一清颗粒制剂的定量测定。

Objective：The quantitative analysis model for determination of the content of baicalin in Yiqing granules of 85 samples from 25 manufacturers was established using near - infrared diffuse reflectance spectroscopy （NIRDS） combining with partical least squares （PLS）. Methods：The near infrared diffuse reflectance spectra of the samples were acquired in 12000 -4000 cm^-1. Different wavenumber ranges and spectrum preprocessing methods were investigated according to the root mean square errors of correction （RMSEC） values and the correlation coefficient（R2）. This model was developed to correlate the spectra and the values determined by HPLC. The model was successfully applied to predict 10 unknown samples. Results： The R2 between true and prediction was 0. 9732. The RMSEC was 0. 0625 and the root mean square errors of prediction （RMSEP） was 1. 5031. The average predication recovery was 101.1%, RSD was 3. 0%. Conclusion：It can be used rapidly and correctly for determination of baicalin in Yiqing granules. It owns many remarkable advantages that can not be displayed by traditional analysis methods, such as nondestruction, no pollution,good reappearance. NIRDS can control the quality of Yiqing granules.