Abstract
The paper proposes a new method for finding the optimal design of complementary split ring resonator cells for a microstrip antenna, which makes it possible to purposefully optimize the parameters of resonator cells, which significantly improve the electrical characteristics of the antenna. This antenna is designed for cellular communication 5G. The proposed method combines a regression model using a deep network based on fully connected neural layers with a search based on the conditional optimization algorithm COBYLA to find the optimal design parameters of resonator cells. This approach fully automates the process of creating microstrip antennas with resonator cells based on metamaterials and makes it possible to find the parameters of complementary split ring resonator cells that provide the optimal operating mode as a whole and gives the maximum level of antenna radiation while maintaining the bandwidth. The small size of complementary split ring resonator cells opens up the possibility of using this peculiar configuration to create compact and efficient microstrip antennas.