Optimization Of Sensor And Research Of Reconstruction Algorithm Of Direct3D Electrical Capacitance Tomography

Electrical Capacitance Tomography(ECT) technology is used to measure thecapacitance value between the electrodes by electrodes uniformly arranged around themeasured object, and using these capacitance values reproduce distribution of differentconstant material inside the measured object by algorithm. Compared with othertechniques, capacitance tomography technology has been widely applied to two-phaseflow measurement in petroleum, chemical, power and metallurgical industries for itsnon-invasive, simple structure, portability, cheap, no radiation hazards, easy to implement,fast response time and other advantages.Direct3D ECT reconstruction is arrang multi-plate measurements around themeasured object and measure the capacitance value of both the same layer and the differentlayer plates.3D imaging generate the image directly from the reconstruction processskipping the error link of two-dimensional tomography. This can get a higher qualityreconstructed image, so it has great research value.In this paper, the author use multi-physics coupling software Comsol Multiphysics todo forward problem modeling of direct3D ECT system, and using Matlab software to doinverse problem reconstruction. This paper will do research on18plate sensor structurewhich has three layers in the axial direction and six plate on each layer. Mainly examinethe affect on direct3D ECT system caused by the sensor shielding structure, the axiallength of the plate and the rotating of adjacent layer plate. By analysis of the maximum andminimum capacitance value, the dynamic range of the capacitance, sensitivityhomogeneity, the condition number of sensitivity matrix and the result of imagereconstruction can obtain the above structure parameters’ impact on direct3D ECT system,and determine the sensor structure has more advantages than others.To improve the reconstruction speed, eliminate the low gray artifacts, obtain betterand faster ECT reconstruction image, this paper proposes a new method to obtain thresholdautomatically and apply it to Landweber algorithm in two ways then compare them with Landweber algorithm and unified threshold Landweber algorithm. The two ways are calledadaptive threshold filtering and embedded threshold filtering. Three kinds of indicators areused to assess the effect of image reconstruction. The simulation result shows that the twonew threshold filtering compared with Lanweber algorithm and unified thresholdLandweber algorithm has significantly improved on both calculation and reconstructedimage quality. In the two new threshold filtering, adaptive threshold filtering has betterreconstructed image quality，and embedded threshold filtering has faster reconstructionspeed.