Mathematical Model And Feature Extraction For Electrical Capacitance Tomography System Of Two-phase Flow

Herbal extraction is very important in the pharmaceutical process. As the extraction of herbs were carried out in closed equipment, so the conventional container extraction methods are difficult to determine the distribution of these materials and solvents, concentration parameters and so on. At present, in the pharmaceutical process judging the leaching of medicine and the amount of solvent mainly depend on the operating experiences of the workers. Inevitably, this will produce a great error, and can't extract the active ingredients of herbs in a maximum limitation, all of these can result in waste of resources and increase the corresponding production cost.Electrical capacitance tomography (ECT) can detect the materials' composition, saturation, and other information in the closed equipment, also combining with computer technology can image the distribution of the materials.This topic studied two contents: establishing the mathematical model of the extraction process and extracting features for the distribution of medicines and solvents, the aspects of the research as follows:1. Combined with the background of this issue, it analysises the detection of multiphase flow in production and research, analyzes the characteristics and parameters of multiphase flows ,also analyzes and summarizes the research status and future trends;2. This subject describes using principal component analysis for samples of high-dimensional data dimensionality reduction and classification of support vector machines. Described the principle of principal component analysis and the identification information extraction, and gives a support vector machine classification of the principles of conducting and learning steps; 3. With the actual situation of the subject, using the finite element method for unit cross-section of pipe split, establishing the corresponding finite element equations. Using the finite element method for getting the capacitance, Gauss theorem is improved, the obtained results is more accurate in this way, and the value will serve as the basis for the subsequent feature extraction;4. Shows the feature extraction using principal component analysis method. And established a suitable model for support vector machines and decision-making functions, and the merits and demerits of the nuclear experiment selected the appropriate function. Finally, a simulation experiment results and the test of time from the identified two aspects of analysis to prove that the principal component analysis and support vector machines combined with each other is a good method for feature recognition.