Antenna Array Pattern Synthesize Based on Convolutional Neural Network

In this paper, an intelligent computing method based on convolutional neural network (CNN) is proposed, and finally the radiation pattern is derived from the given phase distribution. The method is validated on a 30*30 metasurface array and both numerical and experimental results are in good agreement with the numerical results. The performance shows that machines can use deep convolutional neural networks to “learn” the physical principles of electromagnetic waves. The trained neural network can predict the desired directional map in milliseconds, greatly reducing computation time.