目的 以热毒宁注射液（Reduning Injection，RI）为研究对象，应用近红外光谱（near infrared spectroscopy，NIRS）法和折光率法，建立快速检测金银花提取和浓缩工序中间体总固体量的方法。方法 收集RI金银花提取和浓缩工序中间体，采用NIRS法与折光率法分别建立2种中间体总固体量的检测方法，对比2种方法检测结果的准确性。结果 2种中间体的NIRS、折光率与总固体量之间均有强相关性，建立的总固体量定量预测模型的相关系数均大于0.97；模型验证结果显示，金银花浓缩工序中，NIRS模型和折光率模型预测性能相近，相对预测误差均小于5%，表明针对金银花浓缩工序，2种方法均可用于检测总固体量；金银花提取工序中，2种模型预测能力相差较大，NIRS模型预测准确性较高，折光率模型预测准确性较低，表明针对金银花提取工序建立的折光率模型不适用，应选用NIRS法检测总固体量。结论 基于NIRS法与折光率法构建的定量预测模型，实现金银花提取和浓缩工序中间体总固体量的快速检测，为RI生产过程质量监测选择适宜的快速检测技术提供了技术参考。

Objective Taking Reduning Injection (RI, 热毒宁注射液) as the research object, the near-infrared spectroscopy (NIRS) method and the refractive index method were applied to establish a method for rapidly detecting the total solid content in the intermediates of the extraction and concentration processes of Jinyinhua (Lonicerae Japonicae Flos). Methods Intermediates from the extraction and concentration processes of Lonicerae Japonicae Flos in RI were collected. Two detection methods for the total solid content of these intermediates were respectively established using the NIRS method and the refractive index method, and the accuracy of both methods was compared. Results Both NIRS method and refractive index method demonstrated a strong correlation with the total solid content in the two intermediates, with correlation coefficients of the quantitative prediction models exceeding 0.97. The model validation results showed that the NIRS model and refractive index model had similar predictive performance in the concentration process of Lonicerae Japonicae Flos, with relative prediction errors of less than 5%. This indicated that both methods could be used to detect total solid content in this process. In the extraction process of Lonicerae Japonicae Flos, the predictive capabilities of the two models differed significantly, with the NIRS model showing higher prediction accuracy and the refractive index model showing lower prediction accuracy. This indicated that the refractive index model was not suitable for the extraction process of Lonicerae Japonicae Flos, and the NIRS method should be used to detect the total solid content. Conclusion Quantitative prediction models based on NIRS and refractive index methods were constructed for the rapid detection of total solid content in intermediates of the extraction and concentration processes of Lonicerae Japonicae Flos. This study provided a technical reference for selecting an appropriate rapid detection technology for the quality monitoring in the production process of RI.

R283.6

国家长三角科技创新共同体联合攻关项目（2023CSJGG1700）