الفهرس 
Survey
Network DesignOnly 14 pages are availabe for public view
 |  | from | 135 |  |  |
| | | from | 135 | | |
Abstract
This thesis is dealing with an inverse scattering problem, where the electric properties (permittivity and conductivity) of a scatterer is determined from the awareness of the source and scattering data. The supervised artificial neural net (ANN) method is chosen to solve the presented problem. The ANN needs a massive number of computed examples for NN’s training. Therefore, a fast and efficient solver of the forward problem (where the scattering data is determined from the awareness of the source and electric properties of the scatterer) is required.
Hence, a new semi-analytical formulation for the calculation of the scattered electromagnetic field from lossy perfect dielectric infinite cylindrically-shaped scatterer is presented. It involves a volume integral equation on an induced electric polarization current inside the scatterer and a complete orthonormal set of radiating/nonradiating polarization currents formation, via Green’s function technique, for solving Maxwell’s equations. The results of the proposed approach are compared with a number of cases that have known analytical solutions. The comparison is carried out for a large variety of scatterer’s diameters, conductivities, permittivities and source operating frequencies. The comparison showed that the presented scheme is a very precise representation of the analytical one. Hence, the proposed approach proved to be one of the most efficient formulations for solving Maxwell’s equations.
Then, a multilayer perceptron ANN is designed, executed and tested for performance on the inverse presented problem. Design factors involved number of hidden layers, different numbers of neurons per hidden layer and different training methods. The performance results’ analysis proved that multilayer perceptron ANN are effective in solving nontrivial inverse scattering problems; even in the presence of noise.
Finally, an instant (online) solution has been introduced by the multilayer perceptron network in solving the presented forward problem. It proved to be effective in solving nontrivial forward problems in the field of electromagnetic waves’ scattering.