The Study Of The Applications Of Artificial Neural Networks In Capillary Electrophoresis And Process Analysis

Since the research works of Jorgenson and Lukacs were published in 1981, the separation technology by Capillary Electrophoresis(CE) has been developing greatly. High performance, less time for analysis and much less amount of sample are the strongpoints ofCE. In the analysis of organic and drug molecules, especially chiral compounds and biological marcomolecules, CE challenges the status of HPLC. On the other hand, CE has its objections. Its reproducibility and precision of determination are worse than those of HPLC. In order to overcoming the objections of analysis by CE, the possibility of using Artificial Neural Networks(ANNs) to increase the precision of determination by CE is studied. The thesis is divided into 5 chapters. The uses of ANNs in CE and Chromatography are reviewed, and the possibility of using ANNs to increase the precision of determination by CE is studied, and the using of ANNs in Process Analytical Chemistry is also reviewed. The outlines of the main work of the thesis are:1. The progress of the applications of ANNs in CE and Chromatography is reviewed, including prediction of migration or retention behaviors, optimization of the experimental conditions, pattern recognition, quantification of the overlapped peaks, judgement of the peak purity and modeling of nonlinear correspondence, etc.2. A new method using ANNs for the calibration of the determination ofmemantine by CE is proposed. A ' soft model' based on ANNs for determination is created.3. The optimized experimental conditions for the determination of ribavirin by Capillary Zone Electrophoresis(CZE) is sdudied. A new model based on ANNs bet\veen the concentrations of ribavirin and its corresponding responding signals is built. Using the model, satisfied results of determination of ribavirin by CE are acquired.4. The applications of ANNs in the Process Analytical Chemistry are reviewed, including process modeling, process control, pattern recognition and clusteringanalysis and the applications ofANNs in electronic nose.