一种基于近红外光谱技术的熊胆粉粗提物洗涤工艺优化方法

在熊胆粉粗提物的乙酸乙酯洗涤工艺中，需要根据原料的性质对所用洗涤溶剂的倍数进行调整，才能提高产品的纯度和产率。该文采用近红外光谱反映熊胆粉粗提物原料的整体性质，对这一工艺进行了优化。按照不同的熊胆粉粗提物和乙酸乙酯用量设计试验，采集熊胆粉粗提物的近红外光谱，提取出光谱数据的前8个主成分，同乙酸乙酯用量一起作为输入变量，对熊胆粉提取物成品的纯度和收率进行回归。试验共计40组，其中30组校正集，用于建立模型；10组验证集，用于对所建模型进行评估。所建模型的校正、内部交叉验证和外部验证结果满意，纯度模型的相关系数分别为0.902，0.896，0.883，校正误差均方根（RMSEC），交互验证均方根误差（RMSECV），预测误差均方差（RMSEP）分别为1.22%，1.48%，1.59%。收率模型的相关系数分别为0.921，0.859，0.916；RMSEC，RMSECV，RMSEP分别为1.39%，1.65%，1.53%。结果表明，熊胆粉粗提物的近红外光谱结合乙酸乙酯用量可准确预测所得成品的纯度和收率，利用所建模型可以根据熊胆粉粗提物的性质优选出最为合适的乙酸乙酯用量，提高熊胆粉提取物的纯度和收率。

Technological parameter optimization method for washing of coarse bear gall powder extracts based on NIR spectroscopy

During the washing process of coarse bear gall powder extracts, it is necessary to adjust the amount of ethyl acetate according to the properties of raw materials, which aims to improving the yield and purity of the final product. In the research, using NIR spectra to reflect the comprehensive properties of coarse bear gall powder extracts, the process is optimized in a flexible way. Forty batches experiments are designed according to the weight ratio of ethyl acetate and coarse extracts of bear gall powder. The NIR spectra of the coarse extracts of bear gall powder are collected and processed using principal component analysis (PCA) method. The first 8 principal components combined with the amount of the ethyl acetate are used as the input variables, and calibration models are established to predict the yield and purity of the final product. 30 batches are used as calibration set, which is used to establish the models, and other 10 batches are used as validation set, which is used for the performance appraisal of the established models. The correlation coefficients of the calibration, inner cross-validation and external validation for the purity model are 0.902, 0.896 and 0.883, respectively, and the RMSEC, RMSECV and RMSEP are 1.22%, 1.48% and 1.59%, respectively. The correlation coefficients of the calibration, inner cross-validation and external validation for the yield model are 0.921, 0.859 and 0.916, respectively, and the RMSEC, RMSECV and RMSEP are 1.39%, 1.65% and 1.53% respectively. This work demonstrated that NIR spectra combined with technology parameter could be used to predict the yield and purity of the final product. Using the established models, the most appropriate amount of the ethyl acetate can be determined according to the properties of the coarse bear gall powder extracts, and the yield and purity of the final product can be improved.