An Antenna Array Design Method Based on Residual Graph Neural Network

A learning method based on residual graph neural network was proposed to improve the efficiency of antenna array design. Firstly, the geometric parameters of the antenna array are processed into graph representation, which is used as the input features of the graph neural network. After the graph encoding process, the complex-valued fully connected network with residual connections is used to predict the radiation pattern of the antenna array. Based on the residual graph neural network, a seven-element linear array structure is designed to prove its effectiveness.