전기 임피던스 단층 촬영을 위한 전산해석체계 개발

Abstract

The electrical impedance tomography(EIT) is an imaging modality that reconstructs the resistivity or conductivity distribution of an internal object through various image reconstruction algorithms based on measured voltage at the boundary of the domain. This technique has been applied to medical, chemical processing, prediction of two-phase flow behavior and others. However, there are mainly two difficulties in analysis of the EIT as followings. The first thing, there are limitations of the application to realistic problems due to difficulties in generation of complicated geometry. The second thing, mesh generation and arrangement is very difficult, and in case of generating the multitudinous number of the element the generation time is quite much required. On the other hand, it is possible to conveniently describe complicated configurations and to easily generate the mesh if using commercial softwares. In this study, unified algorithm which can be applied to the EIT technique was developed by linking COMSOL Multiphysics based on finite element method and a self-developed computational analysis system in order to overcome these problems. At first, for the verification of the reliability of the computational analysis system this study developed the forward solution was compared with the exact solution, that of EIDORS which has been widely used as a analysis package for the EIT, and of the boundary element method. For the assessment of the easiness for two difficulties which were mentioned before, at first comparisons of the mesh generation time with that of the QMG which has been widely used, and also the validity of the proposed computational analysis system for the complicated geometry wasevaluated considering a geometry that has turbine blades inside the two-phase flow. The developed computational analysis system was applied to the static and dynamic problem of the EIT technique. That is, various image reconstruction results for static problems were obtainedapplying well-known inverse algorithms such as Gauss-Newton(GN), Conjugate Gradient Method(CGM), Particle Swam Optimization(PSO), Mesh Grouping Method(MGM) and others to the proposed computational analysis system. Similarly, for the image reconstruction for dynamic problems the analysis for the bubble motion was conducted applying the Extended Kalman Filter (EKF) to the proposed system. This study largely consists of four parts as followings. In the first chapter, the brief introduction of EIT technique, in the second chapter, the mathematical background for demonstration of EIT technique are dealt with. In the third chapter, technical parts and verified results for the proposed computational analysis system are presented, and in the fourth chapter reconstructed images for static and dynamic problems obtained by applying various inverse algorithms to the proposed system are introduced.