Authors: Bilel Hamdi, Taoufi Aguili
Source: FERMAT, Volume 22, Communication 3, Jul.-Aug., 2017
Abstract: Studying of mechanical steerable antennas has been considered the subject of feature research. In order to reduce the cost of the mechanical system and to grow up the steering capabilities of the radar, we suggest replacing any mechanical antenna components with an electronically controlled 3D or conformal antennas arrays. 3D antenna arrays can be easily produced with existing manufacturing technologies and offer a considerable advantages in terms of 3-D steerable radiation beam, size, directivity, HPBW and SLL reduction. In this paper, we have developed the neural networks method based beamforming that will be applied to the array pattern synthesis for three-dimensional (3D) conformal antenna arrays. This approach permits to model and optimize the antenna arrays system, by acting on many parameters of the array and taking into account predetermined general criteria. The goal is then to build a feed-forward neural network with supervised learning that approximates the following array pattern’s function. It explains how to introduce the basic principles of artificial neural network (ANN), some fundamental networks are examined in detail for their ability to solve simple pattern synthesis problem in conformal antenna arrays. This fact increases the complexity of the problem under consideration and fitting the neural network model, such as training function, architecture and parameter that would improve and result more accuracy about input-output relations. Then, the used neural technique proved its effectiveness in improving performance using the known conformal isotropic antenna arrays.
Index Terms: mechanical steerable antennas
View PDF
Optimization of Electronically Scanned Conformal Antenna Array Synthesis Using Artificial Neural Network Algorithm