【作者】 李斌；

【导师】 瞿海斌；

【作者基本信息】 浙江大学 ， 生物化工， 2005， 硕士

【摘要】 我国中药的生产工艺处于比较落后的状态,其生产过程主要凭经验来控制,缺乏有效的在线监测技术,直接影响中药制造过程的质量,进而影响中成药的最终质量。近红外光谱技术作为一种快速、无损的过程分析方法,已在众多工业领域的过程分析和质量控制中得到应用,但在中药制药过程的应用研究却很少。本文以中药制药过程为对象,研究了近红外光谱技术在中药制药过程在线质量分析中的应用。主要工作有: 1.以红参乙醇提取过程为例,研究了醇提过程的近红外光谱快速分析方法。在提取过程中采集提取液样品,扫描样品近红外光谱,用比色法测定样品的总皂苷浓度作为对照值。运用正交信号校正算法除去光谱中的干扰信息,然后采用偏最小二乘回归算法建立近红外光谱校正模型。校正集和验证集的预测均方差分别是0.15mg/mL、0.16mg/mL,相关系数均为0.99。研究结果表明,利用近红外光谱技术能够准确、快速地测量红参提取液的总皂苷浓度,可用于快速分析中药醇提过程。 2.建立了红参醇提过程的近红外光谱在线分析方法。在线采集提取液的近红外光谱,以比色法测定的提取液中总皂苷浓度作为对照值。标准正则变换(SNV)结合一阶导数对光谱进行预处理后,采用偏最小二乘回归算法建立在线测量总皂苷浓度的近红外光谱校正模型。校正集和验证集的预测均方差分别是0.15mg/mL、0.17mg/mL,相关系数分别是0.99、0.98。 3.丹参水提生产过程的近红外光谱在线分析方法的研究。在丹参工业生产过程中在线采集提取液的近红外光谱,以高效液相色谱法测定的提取液中丹参素和原儿茶醛浓度作为对照值,采用偏最小二乘回归算法建立近红外光谱校正模型。丹参素和原儿茶醛的预测均方差分别是0.0282mg/mL、0.00448mg/mL,相关系数分别是0.87、0.88。 4.建立了红参醇提液浓缩过程的近红外光谱在线分析方法。用无水乙醇按比例稀释红参浓缩液配制标准样品,获得其乙醇浓度和总皂苷浓度的对照值和近红外光谱,建立近红外光谱与对照值之间的校正模型,然后使用该模型在线分析红参醇提液的浓缩过程。总皂苷和乙醇的预测均方差分别是1.81mg/mL、1.58%,相关系数分别是0.98、0.99。 5.研究了红参浓缩液脱色过程的近红外光谱快速分析方法。采用正交信号校正算法预处理近红外光谱,以遗传算法作为波长选择方法,用偏最小二乘回归算法建立近红外光谱校正模型。总皂苷和透光率的预测均方差分别是0.96mg/mL、1.13%。 上述研究工作表明,近红外光谱技术可用于中药制药过程的在线质量分析,可进一步发更多还原

【Abstract】 Technics of Traditional Chinese Medicine (TCM) manufacturing process is under-developed, and process has been monitored with experience. The absence of effective on-line monitoring techniques will affect the quality of TCM manufacturing process directly, and the quality of TCM farther. As a fast, non-destructive process analysis method, Near Infrared Spectroscopy (NIRS) technique has been applied in fields of industry, but little in TCM manufacturing process. In this thesis, the application of NIRS technique in on-line quality analysis of TCM manufacturing process had been researched. The main works can be summarized as follow.1. Taken ethanol extraction process of Red Ginseng as an example, the method of fast analysis of ethanol extraction process by NIRS had been investigated. Samples were collected during extraction process, and then the spectra of samples were scanned. The gensenosides concentrations of samples were measured by colorimetric assays as reference values. The interference information in the spectra was deleted by orthogonal signal correction method. A calibration model between spectra and reference values was built by partial least squares regression method. The root mean square error of calibration set and validation set were 0.15mg/mL, 0.16mg/mL, and correlation coefficient were all 0.99. The results showed that the predictive accuracy of NIRS calibration model used for determination of ginsenosides concentrations was good.2. The method of on-line analysis of ethanol extraction process of Red Ginseng by NIRS had been built. Spectra were on-line collected with flowcell. Gensenosides concentrations of samples were measured by colorimetric assays as reference values. First derivative combined with standard normal variate was used to pretreat spectra, and calibration model for ginsenosides had been built by partial least squares regression method. The root mean square error of calibration set and validation set were 0.15mg/mL, 0.17mg/mL, and correlation coefficient were 0.99, 0.98.3. The method of on-line analysis of industrial water extraction process of Radix Salviae Miltiorrhizae by NIRS had been studied. Spectra of extract were obtained with industrial flowcell in the course of manufacturing process. Partial least squares regression method was used to build the calibration models with HPLC as a reference method for each component. Root mean square error of prediction for Danshensu and salvianolic acid B were 0.0282mg/mL,0.00448mg/mL,correlation coefficient were 0.87, 0.88 <>4. NIRS was used in on-line analysis of concentration process of Red Ginseng alcohol extract. The standard samples were prepared by diluting concentrated Red Ginseng extract proportionally with anhydrous ethanol, and their reference measurements of alcohol and ginsenosides concentration and spectra were obtained. The calibration models for alcohol and ginsenosides were built. Models had root mean square error of prediction of 1.81 mg/mL, 1.58%, and correlation coefficient of 0.983, 0.997 for ginsenosides and alcohol.5. NIRS had been explored for fast analysis of decolourisation process of Red Ginseng concentrated extract. Spectra were pretreated with orthogonal signal correction method, and wavelength was selected by genetic algorithm. Calibration models were developed by partial least squares regression to measure gensenosides concentrations and transparency. Root mean square error of prediction for gensenosides and transparency were 0.96 mg/mL, 1.13%.As those works shown, the NIRS technique could be applied in on-line quality analysis of TCM manufacturing process, and developed to be a new method of monitoring the quality of TCM manufacturing process.更多还原