目的 建立青蒿单效浓缩过程中近红外在线快速检测模型,并讨论吸光度的变化对其模型建立的影响。方法 在线收集9批浓缩液样本,偏最小二乘(PLS)法建立定量校正模型,并用此模型对1批样品进行预测。结果 总酸和固含量PLS模型参数如下:决定系数(R²)分别为0.9679和0.9623,校正集均方根误差(RMSEC)分别为0.7835和0.9488,交叉验证集均方根误差(RMSECV)分别为0.8258和0.9780。结论 青蒿浓缩液样品吸光度范围为0~2.0,该光谱建立的总酸和固含量的PLS模型的预测相对偏差(RSEP)值均在10%以内,能够满足青蒿生产过程中质量要求,说明当吸光度范围为0~2.0时,通过增加样品数和运用化学计量学方法可消除吸光度太高(即透过率太低)对青蒿近红外模型建立的影响。

Objective In this study, an in-line near infrared spectroscopy (NIRS) rapid detection model of single-effect concentration process of Artemisiae Annuae Herba (AAH) was established and the impact of absorbance to model was discussed. Methods Nine batches of concentration process samples were collected inline and partial least square (PLS) method was applied to build its quantitative model which was used to predict an unknown batch of samples. Results The parameters of PLS model of total acid and soluble solid content were as follows:coefficient of determination (R²) of 0.9623 and 0.9679, root mean square errors of calibration (RMSEC) of 0.7835 and 0.9488, and root mean square errors of cross-validation (RMSECV) of 0.8258 and 0.9780, respectively. Conclusion Absorbance of concentration process samples of AAH was between 0 and 2.0 and RSEP values of total acid and soluble solid content was less than 10%. It can meet the quality requirements of production process of AAH and indicate that the effect of high absorbance (namely low tansmittance) on NIRS model of AAH can be eliminated by adding samples and using chemometrics when the absorbance is between 0.0 and 2.0.

科技部重大新药创制现代中药创新集群与数字制药技术平台(2013ZX09402203)