Study On Image Reconstruction And Flow Pattern Identification For Electrical Capacitance Tomography System

Electrical capacitance tomography (ECT) is a novel technology of computed tomography imaging technology, which brought forward by UMIST of U.K. in the late 1980s. It has a number of capacitive electrodes mounted around the periphery of measured object, and measures the capacitance value between electrodes to get the distribution of dielectric constant, which is mainly used in detection of multi-phase flow in industrial pipeline. It can provide visualized information of the concentration, distribution and status of movement of multiphase pipelines. Compared with the other technologies, electrical capacitance tomography technology has obvious advantages, such as wider range of uses, non-invasive and better safety performance, and is applicable to a wide range of industrial production process of common multiphase flow detection, and costs less. More profound study is focused on the key problems such as image reconstruction and flow pattern identification. By study on 12 electrodes electrical capacitance tomography system we have done is as follows.Principle analysis of electrical capacitance tomography system. From the the oretical analysis of electrical capacitance tomography technology works, the establishment of a capacitance-sensitive field of the math-ematical model and use it as the image reconstruction and flow pattern recognition theory.Analysis of image reconstruction algorithms. Image reconstruction algorithm for ECT are mainly linear back projection algorithm, model-based MOR method, iterative algebraic reconstruction method, look-up table method, neural network method and Tikhonov methods brief introduction and comparative analysis. By Tikhonov regularization analysis, the standard Tikhonov functionals over smooth, resulting in the reconstruction of the details of image information is missing, the reconstruction of the image quality problem is not ideal. In this paper, using a standard algorithm based on Tikhonov, a new iteration operator is given, using the operator can make the reconstruction of the image detail for certain amendments, through simulation experiments prove that at the speed and accuracy of image reconstruction has been improve.Flow pattern identification analysis and improvement. Several major flow pattern identification methods are introduced: nearest neighbor method, K nearest neighbor, neural network and feature extraction method. And their advantages and disadvantages are analyzed and compared. The previous flow pattern identification method takes small number of pixel data to identify the flow pattern. By increase the number of pixel data to improve recognition accuracy rate.Simulation results proved that more accuracy rate of flow pattern identification rate is improved by this method.