R.M.S. de Oliveira, R. Chamié, R. de C. Rocha, W.H. Barros Jr., C.L.S.S. Sobrinho, and R.C. Leite (Brazil)
Parallel Computational Environment, LN-FDTD method,LN-UPML technique, Neural Networks, GeneticAlgorithms, Particle Swarm Optimization.
The present work shows the results of a R&D project carried out by ELETRONORTE and by the Federal University of Pará (UFPA). Its core objective is the development of a computational system for performing analysis and synthesis involving electromagnetic interference (EMI) in a power system substation environment. For the analysis stage, the numerical solution of the problem is obtained by solving numerically the Maxwell’s equations written in a local non-orthogonal coordinate system by employing the Finite-Difference Time-Domain Method (LN-FDTD). The truncation of the analysis region is done by a new formulation called LN-UPML which involves the solution of the Maxwell Equations for lossy anisotropic media in a general coordinate system. For the synthesis stage, techniques such as neural networks, genetic algorithms and particle swarm optimization are used. The computational environment conceived has a friendly data input/output interface for the user, which permits a better understanding of the electromagnetic phenomena involved in each one of the complex problems analyzed.
Important Links:
Go Back
