Abstract

The method for the constructive synthesis of planar antenna arrays (AAs) based on the neural network approach with consideration of interelement coupling is proposed. Within the AA mathematical model, it is assumed that when substantiating the AA configuration, antenna elements can be divided into groups of active, unused, or passive elements with the currents induced by neighbor elements. The structure of a neural network including a classifying neural network and several approximating networks is substantiated. The learning algorithm for the neural network is divided into two stages: the stage of a separate pre-tuning of each part of the neural network and the stage of joint training of all its parts. We have obtained the solution to the problem of the constructive synthesis of AA according to the requirements for the shape of the radiation pattern. The solution confirms the method workability and good prospects of the neural network approach.

中文翻译：

---

考虑元件间耦合的平面天线阵列结构综合方法

摘要

提出了一种基于单元间耦合的神经网络方法构造平面天线阵列的方法。在AA数学模型中，假设在证实AA配置时，可以将天线元件分为有源元件，未使用元件或无源元件组，并通过相邻元件感应出电流。证实了包括分类神经网络和几个近似网络的神经网络的结构。神经网络的学习算法分为两个阶段：神经网络各部分的单独预调整阶段和所有部分的联合训练阶段。根据辐射图的形状要求，我们已经获得了AA的结构合成问题的解决方案。该解决方案证实了该方法的可操作性和神经网络方法的良好前景。