论文部分内容阅读
采用小试近红外(NIR)分析模型监测大孔树脂纯化栀子提取物的放大过程。首先,收集小试纯化过程中的乙醇洗脱液,采集其近红外光谱,采用HPLC测定栀子苷含量,并采用偏最小二乘法(PLS)建立NIR光谱预测栀子苷含量的定量模型。然后,用该模型对中试过程醇洗脱液中栀子苷浓度的变化进行监测。结果表明,小试NIR模型对中试过程中栀子苷浓度的预测效果良好,然而随着实验批次的进行,模型的预测性能有所下降,因此用模型更新的方法对该模型进行维护。经过2次更新后,模型可以对中试过程中栀子苷的浓度进行准确预测。通过模型更新将小试规模建立的NIR定量模型应用于不同规模的大孔树脂纯化过程,可提高小试过程数据的利用效率,并且节省中试过程重新建立模型的成本。
A small-scale near-infrared (NIR) analytical model was used to monitor the magnification of macroporous resin purified gardenia extract. First of all, we collected the elution liquid of ethanol from small sample and collected its near-infrared spectra. The content of geniposide was determined by HPLC. The quantitative prediction model of geniposide was established by Partial Least Squares (PLS). Then, the model was used to monitor the change of geniposide concentration in the alcohol elution solution of the pilot process. The results showed that the prediction accuracy of jasminoidin concentration in the pilot plant was good with the pilot NIR model. However, as the experimental batch progressed, the predictive performance of the model declined. Therefore, the model was updated by using the model updating method. After two updates, the model can accurately predict the concentration of geniposide in the pilot process. Through the model updating, the NIR quantitative model established in the pilot scale can be applied to different sizes of macroporous resin purification process, which can improve the utilization efficiency of the pilot process data and save the cost of the pilot process to rebuild the model.