Dimensionality reduced simultaneous iterative reconstruction technique for electrical resistance tomography

Abstract

With the advantages of non-invasion, non-radiation, low-cost and high-speed, electrical resistance tomography (ERT) is suitable for online measurement of multiphase flow inside a pipeline. However, due to the highly non-linear and ill-posed characters of the inverse problem, its spatial resolution is low. As a kind of promising solutions for the issue, the iterative methods are always with low computational efficiency. In the paper, a Dimensional Reduction Simultaneous Iterative Reconstruction technique (DR-SIRT) is proposed to solve the two and three dimensional ERT problems. Scalar of the algebraic system for the solution of the inverse problem is significantly reduced by the orthogonal projection operator. The over-smoothing character of the iterative regularization method is mitigated by the non-negative constraint. The results from the numerical testes and the statistic experiments proved that the usage of the DR-SIRT can save the computational resources both in 2D and 3D cases. The discussions on the factors influencing the imaging speedy and quality, such as the number of electrodes and the truncation coefficient in the projection operator, provide some guidelines for the practical applications of the DR-SIRT algorithm.