Study On Identification Of Flow Regimes Based On Feature Extraction And Neural Networks For Electrical Capacitance Tomography System

Two-phase recognition technology has a high application value in the industry, and its development prospects can't be overlooked.With the rapid development of computer technology and signal-recognition technology, Electrical Capacitance Tomography is developed in a very high speed in recent year. Electrical Capacitance Tomography (ECT) technology is a kind of Process Tomography which is based on capacitance-sensitive mechanism, and it has the advantage of non-radioactive, non-invasive, fast response, simple structure, low cost, wide application, security, good performance etc.Flow pattern recognition technology is an important technology in the electrical capacitance tomography technology.The research object in this paper is 12-electrode capacitance tomography system and oil/water two-phase flow,a new flow pattern characteristic parameter is sented after study a large number of ECT Flow regime identification methodology and analysis of experimental data,and these characteristic parameters are analyzed and compared.Meanwhile,it is combinated with the neural network technology,and the model of network technology is established,trained and simulated in Matlab.In this paper,two neural network model, namely the BP neural network and RBF neural network are established.The experiment of six kinds of classical flow patterns of two-phase are done by modeling and simulating in this paper.The experimental results show that by using the characteristic parameters of feature extraction in neural network input, the efficiency and accuracy of the flow-based are better. This paper presents a two-phase flow pattern identification of research and analysis methods which will make the two-phase flow pattern recognition theory, a certain development, but also will be able to as soon as possible capacitance tomography technique widely used in industry to help provide an enabling.