ET Clustering Imaging Algorithm Based On Sensitivity Coefficient

Electrical tomography technology (ET) is a new kind of process tomographytechnology, although this technology has many advantages compared to otherimaging techniques, such as the no radiation, noninvasive, low cost, etc. This makes iteasier in the laboratory research and extension. In addition, it also has thecharacteristics of portability, and visualization.The accuracy and stability of electrical tomography has not reached therequirements of industrial and medical field. Electrical tomography algorithm is ofdiversity, but these algorithms are also have some limitations such as failed to obtainideal imaging effect. One main reason is that the electrical tomography system hassoft field effect, as well as the ill-posed problem for solving nonlinear equations andthe equation of underdetermined problem. Aimed at these problems we put forwardsome methods in this paper, ET clustering imaging algorithm based on sensitivitycoefficient, Tikhonov regularization method, and some data preprocessing methods.In this paper, I completed the following work:Converts16electrodes model to48electrodes model, introducing the sensitivitycoefficient to imaging algorithm that is to say this will reflect the system moreactually, the sensitivity matrix reflect soft field effect. Simulate the system inCOMSOL Multiphysics3.4.Adding a prior information to Tikhonov regularization method, and use thismethod to simulate the process of lungs. Different from the existing methods, theproposed method does not change the relatively weights to any possible solutionavailable, and therefore results in an optimal solution with stable and generalcharacteristics.Spectral clustering method was applied to clustering process, and I put forwardthe data preprocessing method with kalman filtering method and the data evolutionmethod.