Planar array synthesis with scattered inverted-element locations using neural networks

Abstract

A new approach is proposed for the planar array synthesis problem. The method assumes the array elements to be scattered in a plane with each element having a brother element with the same excitation but at an inverted location. These assumptions allow the analogy between the array factor and the output of an artificial neural network (ANN) that is made to learn it. Thus, the parameters of the planar array can be extracted from the weights and biases of the trained ANN. Results of using the approach to synthesize uniformly spaced nonuniformly excited arrays and nonuniformly spaced nonuniformly excited arrays are shown and compared.