Particle Swarm Algorithm Applied to Image Reconstruction on Multiphase Flows
B. C. B. N. Souza1, V. Y. S. L. Cavalcanti1*, M. B. Soares1, F. A. Belo1, G. F. Basso2 and R. A. M. Junior3
1 Laboratório de Energia Solar, LES/CEAR-UFPB
2 Centro de Informática, CI-UFPB
3 Departamento de Energias Alternativas Renováveis, CEAR-UFPB
UFPB Campus I, Cidade Universitária, S/N - João Pessoa, Paraíba, Brazil CEP: 58.051-900
This work presents a methodology for multiphase flow electric capacitive tomography (ECT)  image reconstruction, employing an particle swarm optimization (PSO) algorithm [2,3] in the paradigm of parallel processing. As a general objective, we aim at improving the efficiency of the inverse problem algorithm employed at ECT, in order to increase the spatial resolution of the reconstructed images, without necessarily increase the total processing time of the image reconstruction technique. One important limitation toward this goal is that, for reconstruction techniques of the inverse problem type for ECT, the sensor system response is nonlinear, and as such the total processing time grows faster than any increase in resolution, therefore imposing a high computational cost for real time applications . A first contribution is made by removing unnecessary and redundant processing from the usual code, increasing the overall efficiency of the algorithm. Second, we construct a PSO algorithm for the image reconstruction, demonstrating its effectiveness and efficiency. Then, we implement our routine in a parallel processing fashion. We also present the physical principles of the ECT, the heuristic algorithms used in the reconstruction process as well as the main concepts for parallel computing.
Keywords Electrical capacitive tomography, finite elements, inverse problems, particle swarm optimization, parallel processing, multiphase systems.
Copyright © International Society for Industrial Process Tomography, 2016. All rights reserved.