Particle Swarm Algorithm Applied to Image Reconstruction on Multiphase Flows

Abstract

This work presents a methodology for reconstructing multiphase flow electrical capacitive tomography (ECT) images, using a particle swarm optimization (PSO) algorithm in the parallel processing paradigm. Intended is to improve the efficiency of the inverse problem algorithm in ECT, increasing the resolution of the reconstructed images, without necessarily increasing the processing time of these reconstruction technique. A limitation found is that, for inverse problem-type reconstruction techniques for ECT, the response of the sensor system is non-linear and, therefore, the processing time grows faster than any increase in resolution, imposing a high computational cost. For real-time applications, the first contribution is the removal of unnecessary processing from the usual code; the second is the creation of a new PSO algorithm for image reconstruction that is more efficient than normal. The new parallel processing routine present the physical principles of ECT, the heuristic algorithms used in the reconstruction process and the main concepts for parallel computing.