论文部分内容阅读
目的:采用近红外光谱技术建立氯化铵溶液中氯化铵含量的在线分析方法,并探讨气泡对于检测结果的影响,从而指导中药活性成分的在线检测。方法:以氯化铵水溶液为载体,采集近红外光谱,以偏最小二乘法(PLS)建立模型,基于残差剔除离群值,进行模型优化,通过人工制造气泡考察对模型的影响,并对未知样品进行预测。结果:近红外测量值与实际值相近,预测效果良好,所建氯化铵定量校正模型的相关系数(R2)、内部交叉验证均方差(RMSECV)分别为0.991,0.493。经外部验证,模型的预测均方差(RMSEP)为0.222。当氯化铵浓度大于1.504 g.L-1时,预测相对误差(RESP)控制在10%以内。结果表明气泡对于模型影响不大。结论:采用近红外光谱技术建立的在线分析模型,预测结果的相对偏差满足中药活性成分在线检测的要求,为近红外光谱技术应用于药物在线生产与分析提供了有效的方法和依据。
OBJECTIVE: To establish an on-line analytical method of ammonium chloride content in ammonium chloride solution by near-infrared spectroscopy, and to explore the influence of bubbles on the test results, so as to guide the on-line detection of the active ingredients in traditional Chinese medicine. Methods: Ammonium chloride aqueous solution was used as a carrier to collect near-infrared spectroscopy. Partial least squares (PLS) was used to establish a model. Outliers were removed based on the residuals, and the model was optimized. The effects of bubble on the model were investigated. Unknown samples are predicted. Results: The near-infrared measurement value is close to the actual value, and the prediction effect is good. The correlation coefficient (R2) and the internal cross-validation mean square error (RMSECV) of the ammonium chloride quantitative calibration model are 0.991 and 0.493, respectively. Externally validated, the model has a prediction of mean square error (RMSEP) of 0.222. When the ammonium chloride concentration is greater than 1.504 g.L-1, the prediction relative error (RESP) is controlled within 10%. The results show that the bubble has little effect on the model. Conclusion: The on-line analytical model established by near-infrared spectroscopy can meet the requirement of on-line detection of active ingredients of Chinese traditional medicine and provide an effective method and basis for the application of near-infrared spectroscopy in on-line drug production and analysis.