A Fast Forward Solver for 3D Gain Pattern of Microstrip Antenna Arrays Based on CNNs

A fast forward solver for 3D gain pattern of microstrip antenna arrays based on convolutional neural networks (CNNs) is proposed in this paper. The proposed method utilizes a deep learning approach to predict the gain pattern of the antenna array with high accuracy and efficiency. Firstly, the architecture of the proposed method is introduced. Then the off-line training process of the proposed CNN is described. Finally, the experimental results showcase the on-line performance of the proposed method. The proposed method requires 0.13 seconds for predictions, which is 108 times faster than HFSS, while utilizing only 1/35 of the memory required by HFSS.