论文部分内容阅读
将近红外光谱技术和化学计量学相结合快速检测苯磺酸氨氯地平片辅料含量.通过随机青蛙法、变量投影重要性和竞争自适应重加权采样筛选特征波长变量点,采用9种光谱预处理方法对原始光谱进行处理后,分别建立了偏最小二乘法模型和支持向量回归分析模型,并将这两种模型进行了比较.应用优选模型对样品进行了测试,结果表明:对于所涉及的样本,在最优特征波长变量选择上,随机青蛙法效果较好;在模型预测结果上,与支持向量回归分析模型相比,5个指标的偏最小二乘法定量模型的决定系数,预测均方根误差评价参数效果较好,相对分析误差值均大于3.0.样品测试值与实测值标准误差均小于1.30,配对t检验表明,在a=0.05显著性水平上,两者无显著性差异.因此,可采用近红外漫反射光谱法用于苯磺酸氨氯地平片辅料含量的快速检测,该方法重复性、中间精密度、线性、精确性良好,且可为其他药用辅料含量快速检测提供借鉴.
The rapid determination of amlodipine besylate tablets by the combination of near infrared spectroscopy and chemometrics was used to screen the characteristic wavelength variable points through the random frog method, the importance of variable projection and the competitive adaptive weighted sampling. Nine kinds of spectral pretreatment Methods After the original spectra were processed, partial least square method and support vector regression analysis model were established and the two models were compared.Using the optimal model to test the samples, the results showed that: for the sample involved , The random frog method is better than the random frog method in the selection of the optimal characteristic wavelength. Compared with the support vector regression analysis model, the partial least square method of the five indicators can determine the coefficient of determination, the root mean square of prediction Error evaluation parameters are better, and the relative analysis error values are greater than 3.0. The standard error between the test value and the measured value is less than 1.30, Paired t-test showed that there is no significant difference at the level of a = 0.05.Therefore, Near-infrared diffuse reflectance spectroscopy can be used for rapid detection of amlodipine besylate tablets. The method has the advantages of repeatability, intermediate precision, linearity , Good accuracy, and can provide a reference for the rapid detection of other pharmaceutical excipients.