Non-linear modeling of parabolic reflector: Neural network approach

In this paper, we present artifical neural network (ANN) models for the cal culation of the focal region fields of a parabolic reflector and the estima tion of the direction of arrival. We choose a feedforward NN architecture ( specifically multilayer perceptrons (MLPs)) for their robust and massively interconnected structure. Simulated results achieved with these models show that the neural network approach is a promising alternative to the convent ional methods for focal region calculations of the parabolic reflectors. Th is paper also represents the results obtained from a NN model trained for t he estimation of the direction of arrival for the array-fed parabolic refle ctor.