Neural network approach to shape reconstruction of a conducting cylinder

Abstract

In this study, the shape reconstruction of a conducting cylinder by making use of electromagneticscattered fields is presented. The problem is treated with the Method of Moment (MoM) and FeedForward Neural Networks (FF-NN) is applied. The shape of the cylinder is represented by a Fourierseries while applying MoM, a matrix equation is obtained whose elements are expressed numericallyby using discrete points. The scattered field data and Fourier coefficients of the cylinder are used fortraining the NN as inputs and outputs respectively. The NN results are compared with exact shapes ofsome conducting cylinders; and good agreements with the original shapes are observed. The effect ofthe noise on the scattering field is also investigated.