Quality Control Methods For The Manufacturing Process Of Traditional Chinese Medicine Based On Process Analytical Technology And Chemometrics

Currently,the research objects of the quality control of traditional Chinese medicine(TCM)are focused on crude drug and preparations,and the quality control of manufacturing process and intermediates is ignored.For a series of problems in process quality control of TCM,based on process analytical technology(PAT)and process chemometrics,this dissertation systematically investigated the following contents:near infrared(NIR)spectroscopy in combination with particle swarm optimization based least square support vector machines(PSO-LS-SVM)for quality assessment of TCM and on-line monitoring of manufacturing process;development of an on-line pretreatment system for NIR detection;design and construction of an on-line liquid chromatographic detecting platform.The main works and achievements are included as follows:1.NIR spectroscopy combined with process chemometrics were used for quality assessment of intermediates and preparations of TCM.A novel PSO-LS-SVM algorithm was developed.In this algorithm,the PSO method was applied to select the LS-SVM hyperparameters,including the regularization and kernel parameters.The developed algorithm possesses the capability to escape from local optima,is easy to implement,and has fewer adjustable parameters.Then the PSO-LS-SVM was applied to determine the important quality indicators in complex TCM samples,including:hydroxysafflor yellow A(HSYA)and soluble solid content(SSC)in Honghua extract;uridine,adenosine,guanosine and HSYA in Honghua injection samples;acteoside in the leak solutions and effluents,which respectively collected during macroporous resin adsorption and elution processes of adhesive rehmannia leaf.The results showed that,compared with the conventional principal component regression(PCR),partial least square regression(PLSR),and feed-forward back-propagation network(BP-ANN)methods,the PSO-LS-SVM algorithm with a radial basis function(RBF)kernel exhibited better fitting results and predictive abilities.Furthermore,a comprehensive method using high performance liquid chromatography(HPLC)(including fingerprint analysis and quantitative determination)and NIR spectroscopy was developed to assess the quality consistency of Honghua injection from different manufacturers.The results indicated that the quality consistency of Honghua injections from different batches,especially from different manufacturers,is significantly different.2.As a process analyzer(PAT tool),NIR spectroscopy was employed for noninvasive and on-line quantitative monitoring of manufacturing processes of TCM.These on-line detections were investigated and realized,that included:paeoniflorin,benzoic acid and SSC during extraction process of red paeony root;ferulic acid and SSC during mixed extraction process of Ligusticum chuanxiong Hort.,Salvia miltiorrhiza Bge.and Angelica sinensis(Oliv.)Diels;danshensu,protocatechuic aldehyde,salvianolic acid B,HSYA and SSC during alcohol precipitation of Danhong injection.Calibration-free methods,such as moving block standard deviation,were used to determine the end point of extraction process of Xuebijing injection,and blending process of Zhongsheng Wan powder.To solve the problem of calibration invalidation in extraction process of Xuebijing injection,model updating was adopted to keep the calibration models usable and accurate.In addition,an on-line pretreatment system for NIR detection was presented and developed,which improves the accuracy,repeatability and stability of the collected NIR spectra,and promotes the application of on-line NIR spectroscopy.3.An on-line liquid chromatographic detecting platform was designed and constructed.In this system on-line and automatically pretreatmenting of TCM samples including filtration and flow rate control were performed,and the pretreated liquid was injected into the HPLC system for analysis by six-way valve switch.The calculated results were processed and real-time displayed through Visual Basic programming technique.The on-line liquid chromatographic detecting platform was applied to on-line quantitative monitoring of sodium aescinate A and sodium aescinate B during cold-soaked extraction of Semen Aesculi.The results showed that the on-line liquid chromatographic detecting platform has satisfactory method repeatability and accuracy,and is suitable for real-time analysis during industry manufacturing process of TCM.