Research On Image Reconstruction Algorithms And Gas/Solid Phase Flow Measurement Based On Electrical Tomography

Electrical tomography is one kind of nonintrusive and visible measurementtechnique.Electrical tomography has wide applications in clinical monitoring andindustrial multi-phase flow measurement for the advantages of being fast in response,nondestructive and low cost.In this thesis, image reconstruction algorithms for electrical tomography,optimum design for electrodes and gas/solid two-phase flow measurement, have beeninvestigated. The main works and results are briefly listed as follows:1. Image reconstruction algorithms are deeply studied. An improvedregularization reconstruction algorithm(HTIK)based on3D model is presented. Ingeneral, the regularization coefficient of iterative Tikhonov algorithm is difficult tochoose. Homotopic mapping is adopted to solve the problem. A quasi-Sigmoidmethod is used to adjust the regularization parameter. Thereafter sensitive field imagecan be reconstructed by deploying the resultant image vector as the initial value ofiteration Tikhonov algorithm. Both simulation and experimental results demonstratethat the proposed method can improve imaging quality obviously.2. In order to improve the image quality of electrical tomography, an algorithmbased on the wavelet fusion method is proposed. Simulations are done by constructingan ECT sensor model using the finite element analysis software COMSOL. Theconjugate gradient and singular value decomposition algorithms are utilized toimplement image reconstruction.Then the wavelet-transform-based image fusionmethod is used for image processing. Both simulation and experimental resultsdemonstrate that the method can improve image quality of ECT obviously.Thensimulations are done by constructing an ERT sensor model using COMSOL. Thewavelet-transform-based image fusion method is used for image processing. Bothsimulation and experimental results demonstrate that the proposed method canimprove image quality of ERT.3. To investigate the effect of the number of electrodes on the quality ofreconstructed images, simulations are done by constructing3D ERT sensors withdifferent number of electrodes,4,8,12,16,20,24and32, using the finite elementanalysis software COMSOL. The Landweber, conjugate gradient and truncated singular value decomposition algorithms are utilized to implement imagereconstruction. The simulation results demonstrate that the3D ERT modeling with12electrodes is the best choice.4. The principal component analysis method is applied to gas/solid two phaseflow measurement. In order to identify simulation images,2D image processingmethod based on the principal component analysis is used. Then using PCA method,raw ECT data are used to identify the slug flow in pneumatic conveying system, andthese are confirmed through images reconstructed by conjugate gradient algorithm.Finally, this thesis presents a novel method of calculating the solid velocity based onPCA using cross correlation of twin-plane ECT data.The capacitance signals capturedfrom up-stream and down-stream electrodes are processed by principal componentanalysis. Then the velocity of coal ash in a horizontal tube can be calculated by thecross-correlation method.