Applying Neural Networks for Predicting Feed Weights of an Antenna Array

Artificial neural networks (ANN) are applied to find appropriate phasing for beam steering of patch antenna arrays. To obtain this goal, an ANN is trained using simulated far-field data. Two loss functions are tested and their performance is compared with a uniform and a non-uniform antenna array. The results show that in most cases both tested loss functions perform well, but are not able to find the best solution in all cases.