| The fundamental purpose of the quality control of the pharmaceutical production process was to ensure the uniformity and stability of the final product. So it was necessary to control every link of the whole material circulation from the receipt of the raw material to the ex-warehouse of the end product. The on-line near infrared spectroscopy technology has been widely used in the quality analysis of preparation production links since it is rapid, non-destructive, reliable and easy to operate. In addition, it was the combination of the technology of near infrared and the idea of real-time monitoring of manufacturing process and based on an on-line analysis platform founded by a whole complete facility and of reliable property. Thus, preliminary studies on the development of the on-line near infrared platform, the assessment and verification of the application of the established platform were carried out aimed at the extraction of traditional Chinese medicine, in order to provide guidance for the application of on-line near infrared spectroscopy technology. The contents of the research were as follows:Firstly, based on the working characteristics of production processes and properties of study objects, an on-line NIR measurement platform was established to monitor TCM extraction process by designing a rational NIR detection method. Since it was difficult to achieve in-situ measurement in this study, a bypass detection system was used. According to the different properties of study objects and goals, the bypass system was adjusted and improved regularly. Moreover, spectral data were used to validate and optimize the performance of detection system.Secondly, influences of analysis results from detection system, such as detection methods, sampling methods, were separated to investigate the impact to the final results, respectively, then to evaluate applicability of on-line detection system. Extraction process of Flos Lonicera Japonica as an example, it was proved that the correlation between different detection methods was good by paired T-test and the Pearson correlation coefficient. The bypass system can be applied as on-line process analyzer to monitor the process of TCM water extraction.With the different spectral acquisition methods, partial least squares (PLS) predicted that the standard error of prediction (SEP) were0.0443,0.0432, respectively. SEP were low and no significant difference, it was illustrated that NIR couldn’t be influenced for flowing fluid in bypass system. The external cycle system could be regarded as a reliable method of on-line collecting NIRs in our research. As the research subjects, gardenia fructus extract was filtrated by different sizes of sieve meshes, and then scanned by the near infrared spectra. After the sample was filtrated by300meshes, quantitative prediction model can reach the ideal prediction result (Root Mean Square Error of Cross Validation (RMSECV) was0.1962, Root Mean Square Error of Prediction (RMSEP) was0.1867, Bias was0.0256). It is recommended that proper size of sieve mesh could improve the performance of prediction models dramatically.Thirdly, the main research was the availability of on-line NIRs model. Reliable NIRs were used to build predictive modeling. Then to select variable, to increase the sample size and other methods, which can improve the predictive ability of the calibration model and robustness.(1) The Flos Lonicera Japonica extraction process as the research object, with on-line collecting NIRs, chlorogenic acid quantitative calibration model was established by PLS, and the predictive ability of the model was verified. The wavelength of800-1900nm was selected to build quantitative calibration model. The correlation coefficient R was higher than0.9900. Standard error of calibration (SEC) and SEP was lower than0.0538,0.0593, respectively. It was indicated that on-line NIR detection technology satisfied requirement of Flos Lonicera Japonica water extraction process.(2) The different batches of Flos Lonicera Japonica material were real time monitored in the extraction process.10design schemes were devised to estimate whether or not different batches could disturb prediction result of the quantitative model. According to a series of chemometric indicators, such as SEC, SECV, SEP, RPD, etc., a conclusion could be reached that PLS models established by different batches of raw herbs from same sources could be used to monitor extraction process, and the accuracy of the prediction didn’t decrease sharply in this environment.(3) Flos Lonicera Japonica extraction process of was used as the subject. By comparison with PLS, the conclusion shown predictive performance of internal partial least squares (iPLS) was better than model established by full-spectrum. In iPLS model, latent variable was6, and R2prc, SEP was0.9801,0.068, respectively, with variable range of1640to1779.5nm. RMSECV was lower than0.101.This study developed an on-line near infrared spectroscopy analysis platform for extraction process of traditional Chinese medicine. It would be used as a experimental station for relevant research of on-line analysis. And it provided a demonstrations for development and application of on-line analysis about TCM manufacturing processes. With off-line simulation tools, to analyse online sampling conditions influenced on the results. That provided methodological reference for evaluation of applicability of on-line analytical platform. The applicability of analytical platform was verified by on-line NIR quantitative model, which reflected the significance of on-line platform. |
PDF Full Text Request