Radiation Performance of Satellite Reflector Antennas Using Neural Networks

This paper discusses the development of a neural network array model for predicting the radiation performance characteristics of the horn fed parabolic reflector and the dipole fed corner satellite antennas. A number of neural networks were developed in order to predict the radiation characteristics for various combinations of the design parameters. The results obtained from the neural network array models were compared to those from a commercial design software and found in close agreement. The proposed method can predict in less time and with minimum computational resources, the performance characteristics of a horn fed parabolic reflector antenna with high accuracy.