论文部分内容阅读
目的建立在线监测注射用头孢唑林钠干燥过程中的水分含量测定方法。方法水分定量模型建模样品采以不同批次、不同干燥时间段五水头孢唑林钠原料为研究对象,结合干燥、引湿实验扩充样品水分范围,用积分球测定近红外漫反射光谱,并采用聚类分析挑选出建模光谱,光谱预处理方法为一阶导数+MSC法,谱段为7502~6499.2、5500.2~3 999.8 cm~(-1),回归方法偏最小二乘法。结果水分定量模型建模样品共计85个,水分含量质量分数为12.15%~16.41%,相关系数(coefficient of determination,R~2)为96.30,交叉验证均方差(root mean square error of cross validation,RMSECV)为0.181。外部验证样品52个,水分含量质量分数为13.28%~16.17%,外部验证均方差(root mean square error of predication,RMSEP)为0.149。模型专属性强,准确度、线性和精密度符合要求。结论建立的运用近红外监测注射用头孢唑林钠干燥过程的水分含量方法是可行的,为实现近红外在线干燥提供了实验基础。
Objective To establish a method for on-line monitoring of moisture content of cefazolin sodium for injection during drying. Methods Moisture quantitative model modeling samples were collected from different batches of different periods of drying cefazolin sodium water as the research object, combined with drying, wetting experiments to expand the sample moisture range, with the determination of near-spherical diffuse reflectance spectroscopy and The modeling spectra were selected by cluster analysis. The first order derivative + MSC method was used for spectral pretreatment. The spectral bands were 7502 ~ 6499.2 and 5500.2 ~ 3999.8 cm ~ (-1), and the regression method was partial least squares. Results A total of 85 water samples were modeled with the moisture content of 12.15% -16.41% and the coefficient of determination (R ~ 2) was 96.30. The root mean square error of cross validation (RMSECV ) Is 0.181. Fifty-two samples were validated, with moisture content of 13.28% -16.17% and root mean square error of predication (RMSEP) of 0.149. Model specificity, accuracy, linearity and precision meet the requirements. Conclusion The established method of monitoring the moisture content of cefazolin sodium for injection by near infrared spectroscopy is feasible and provides the experimental basis for near-infrared online drying.