Research On Electrical Capacitance Tomography Direct 3D Image Reconstruction Technology

Electrical Capacitance Tomography (ECT) is a non-destructive imaging technique that aims at visualization of the permittivity distribution of a dielectric object based on the measured capacitance. Through the appropriate image reconstruction algorithm, it seeks to reconstruct the distribution of medium (permittivity) in the measured object field using the measured capacitances, which are obtained from the sensor arrays laid on the measured pipeline or vessel. This technique has the advantages of low in cost, fast in response, easy to implement, non-invasive measurement, safety in performance and wide range in application, so it has wide application prospects.Direct 3D ECT needs to arrange multi-layer electrodes around the measured pipeline, and measures the capacitances between not only the same layer but also the different layers of electrodes. The 3D images directly generate in the image reconstruction process, not through the two-dimensional tomography with error, 3D reconstruction images of higher quality is obtained . So this technique has great research value.In this paper, finite element simulation software is used for a large number of simulation experiments. The author studies the direct 3D ECT image reconstruction technique from the direct 3D ECT sensor as well as image reconstruction algorithm. Direct 3D ECT imaging based on 16 electrodes is researched. These electrodes are arranged in four planes. Every four electrodes on each plane are rotated 0(?) or 45(?) with reference to the previous. The author compares the two sensor structures from the capacitance value of the empty and full pipe, sensitive analysis and image reconstruction quality, then determines the optimum sensor structure.Furthermore, the author presents ECT forward model established by multiple linear regression method, and also uses the improved Landweber iterative algorithm to solve the ECT inverse problem, and compares with LBP and Landweber algorithm on researching. The result shows that the forward problem model established by multiple linear regression method has lighter ill-conditioned modular matrix, and the reconstruction quality obtained by the new algorithm is better than that obtained by LBP algorithm and Landweber iterative algorithm.