A Generalizing Radiation Pattern Synthesis Method for Conformal Antenna Array Based on Convolutional Neural Network

Abstract

With the increasing complexity of antenna array layouts, the efficient methods for synthesizing the radiation patterns of conformal antenna arrays have been the challenge. In this paper, a generalized method based on convolution neural network is proposed to synthesize the radiation patterns of conformal antenna arrays. Furthermore, the Taylor-based aperture projection method is used to generate the training samples and training labels are designed as some two-channel data matrices. Simulation experiments have been designed with a conformal antenna array, of which the radiation patterns of training set and test set are successfully synthesized. The results show that the trained network has high efficiency and good generalization effect for synthesizing the radiation patterns of conformal antenna array.