An Artificial-Neural-Network-Based Efficient Beamforming Synthesis Method and Its Application to Flat-Top Beamformed Compressed High-Order-Mode Dipoles

Abstract

An efficient beamforming synthesis method is proposed for high-order-mode dipoles using artificial neural networks (ANNs). Beamformed radiation pattern features and antenna parameters are set as the inputs and outputs of an ANN model to expedite antenna design by reducing the complexity and training volume of ANN. The flat-top beamforming of compressed high-order-mode dipoles is used as an example to validate the proposed beamforming synthesis method based on a proposed continuous current source over a high-order-mode dipole with the current distribution determined by designed compression coefficients. Then, the desired compression coefficients are implemented using a meandered structure. The numerical results indicate that the ANN can achieve a training loss of $1.16\times 10^{-4}$ and a testing loss of $1.12\times 10^{-4}$ , effectively accelerating the antenna design process. Lastly, a seventh-order-mode printed dipole is designed, simulated, and measured.