【作者】 孙丽英；

【导师】 杨天鸣；

【作者基本信息】 中南民族大学 ， 分析化学， 2008， 硕士

【摘要】 本文围绕近红外指纹图谱,对中药材性味分类、中药材产地鉴别和中药材定量分析进行了研究。本文共有四章。第一章,主要论述了中药指纹图谱的概念及研究现状,简要介绍了近红外光谱法的原理,特点及数据处理方法。阐述了本文的研究意义、研究内容及论文的特色。在不可能将中药复杂成分都搞清楚的情况下,指纹图谱的作用是反映成份复杂的中药内在质量的均一性和稳定性。近红外光谱技术之所以成为一种快速、高效、适合过程在线分析的有利工具,是由其在测试技术上所独有的特点决定的。加上化学计量学方法在解决光谱信息提取和消除背景干扰方面取得的良好效果,使人们对近红外光谱技术进行了广泛的研究。本文利用近红外指纹图谱对中药材的质量进行了定性与定量分析。第二章,中药的药性和药味与近红外指纹图谱的关系研究,主要利用近红外漫反射光谱结合模式识别方法对19种苦味药材、15种甘味药材和20种辛味药材进行分类,并对其中甘味药材中寒性、温性药材和辛味药材中热性、温性进行分类。将各味中药材粉碎过200目筛后采集近红外光谱,然后用模式识别法将所采集的光谱数据进行处理,建立数学模型,利用留一法对所建立的数学模型进行验证。结果显示,热、温性辛味药材的分类准确率为91.0%,寒性、温性甘味药材的分类准确率为86.7%,苦、甘、辛味药材分类准确率为83.0%,达到了满意的结果。本方法说明利用现代手段对不同性味的中药材进行区分是可行的,并且达到了较高的准确率。第三章,中药材产地鉴别研究,利用近红外漫反射光谱法(NIRDRS)对不同产地的丁香进行鉴别。排除了水分含量的干扰,我们利用模式识别的方法分别将采集的丁香各个产地样品的近红外光谱数据进行处理,建立数学模型,并用三重交叉验证法对所建立的数学模型稳定性进行了考察。结果显示,所建立的数学模型对未知丁香的预测准确率为100%。本方法比传统的显微鉴别方法及其它仪器方法(如:HPLC法、薄层法等)方便迅速、准确可靠。第四章,中药材定量指纹图谱研究,分别利用近红外漫反射和透射技术对中药材中水分含量和银杏叶提取液中总黄酮含量进行测定。在水分测定中,用传统的红外水分测定法进行比较,并利用偏最小二乘法建立数学模型。所建立的模型r=0.99912,可以很好地预测中药材的含水量。利用近红外透射法测定银杏叶提取液中总黄酮的含量,所建立的模型具有很好的回归系数,r=0.98206,校正标准差和预测标准差均合理,近红外光谱法和HPLC法所测得的结果无显著性差异(α=0.05),因此,用近红外光谱法结合PLS回归分析可以快速、准确地预测银杏叶提取液中总黄酮的含量,为中药生产在线监测提供了依据。本文从谱图的整体特征出发,使指纹图谱与中药材各组分特征紧密结合,建立了中药近红外指纹图谱,使指纹图谱的研究更加有针对性,质量标准的制定更加有目的性。利用近红外漫反射指纹图谱结合模式识别法对不同性味特征的中药材进行分类,可以为重新界定中药材的性味特征提供依据,也可以指导新药开发及用药配伍。同时,此法用于中药材的产地鉴别也显示出其独特的优势。利用近红外光谱法结合偏最小二乘法的定量的优势,能准确的对中药材中某一种组分进行定量测定,甚至可同时测定多种组分,为复杂的中药有效成分测定提供了一个快速,简便而可靠的方法。更多还原

【Abstract】 This thesis studied the property and flavor classification of traditional Chinese medicine (TCM), habitat identification of TCM and quantitative analysis of TCM by using near infrared spectrometry (NIRS). There were four chapters.Chapter One mainly discussed the concept and development of TCM fingerprint. The principle, characteristic and data processing methods of near infrared spectrometry were briefly introduced. Then we expounded the significance, content and peculiarity of this thesis. Under the circumstances of the impossibility of determining all the explicated components in TCM, fingerprint can reflect the internal homogenicity and stability of TCM. As a fast, high performance, suitable for process-analysis method, NIRS has its typical advantages. The methods of Chemometrics had obtained good effects on extracting information of spectrum and eliminating interference of background. These resulted in the extensive studies on the technique of NIRS. This thesis studied the qualitative and quantitative analysis of TCM by fingerprint of NIRS.In Chapter Two, the property and flavor fingerprint was studied by Near Infrared Diffuse Reflectance Spectroscopic (NIRDRS). 19herbs bitter in flavor, 15 herbs sweet in flavor and 20 herbs pungent in flavor were classified by pattern recognization. Cold and warm herbs in property in the 15 sweet herbs in flavor and hot and warm herbs in property in the 20 pungent herbs in flavor were classified in the same time. All the herbs were gathered their spectra data after pulverized. Then these data was managed by pattern recognization and the mathematic models were established .Then, the models were validated by leave-one-cross-validation (LOOCV) method. The results were that the accurate rate of classification was high to 91.0%(classification of hot and warm herbs in property, pungent in flavor), 86.7%(classification of cold and warm herbs in property, sweet in flavor), 83.0%(classification of bitter, sweet, pungent herbs in flavor) respectively. This method suggested that it was feasible to classifying different properties and flavors of TCM by modern device and the models achieved high accurate ratio.Chapter Three was about the research of the habitat identification of TCM. Flos Caryophylli from different production area was studied by NIRDRS. Eliminated the interference of water, the habitats of Flos Caryophylli were identified by pattern recognition, and the stability of the mathematic models were investigated by cross validation. The results were that the the mathematic models’forecasting correct rate for unknown Flos Caryophylli was 100%. This method is more reliable, accurate, fast and convenient compared with traditional micro-identification method and other equipment method (such as HPLC, TLC et al.).Chapter Four was the study of quantitative fingerprint of TCM, moisture content of TCM and total flavonoids in ginkgo leaves extract were detected by NIRDRS and NIR transmission method respectively. When determined moisture content, NIRDRS was compared with Infrared aquametry, and established a mathematic model by Partial Least Square method (PLS). The result was r=0.99912, and this model can well forecast moisture content of TCM. Total flavonoids in ginkgo leaves extract was determined by NIR transmission method. The coefficient of model was good, r=0.98206. Root mean square error of calibration (RMSEC) and Root mean square error of prediction (RMSEP) were reasonable, and there were no significant differences in the resunlts of HPLC and NIR method(α=0.05). Therefore, NIR combined with PLS can well predict the content of total flavonoids in ginkgo leaves extract.Considered spectra-effect relationship from the whole spectrogram, fingerprint and drug action were combined tightly, multi-dimensional fingerprints of TCM was established, that can better give direction to the study of fingerprint and the institution of quality specification. Meanwhile, this method also showed its typical advantages on the habitat identification of TCM. Utilized the advantage of the combination of NIR and PLS, quantitative fingerprint of TCM was established, that can exactly determine some component in TCM quantitatively. This provides a fast, convenient and reliable method for the determination of the complicated active component of TCM.更多还原