Terahertz Device Design Method Based on Swarm Intelligence Algorithm

Efficient design of terahertz devices that meet technical specifications is a prerequisite for building terahertz communication systems for high-capacity communications. The optimal design of conventional terahertz devices requires manual tuning, which makes the whole optimization process extremely difficult due to the complexity challenges such as the high-dimensional optimization space and the high non-linearity between the decision space and the objective space. To address a series of challenges, we propose a design approach that enables the tuning of terahertz devices through autonomous design utilizing artificial intelligence (AI) techniques, thereby reducing the difficulty of device tuning and improving design efficiency. To validate the effectiveness of the algorithm, we apply it to a microstrip waveguide probe transition structure in the terahertz band. The simulation results indicate that the probe structure designed with the algorithm has S 11 below −20dB in the output frequency range of 291.6GHz ~ 368.6GHz, with a bandwidth of about 77GHz. in the output frequency range of 311.7GHz ~ 367.7GHz, S11 is below -30dB, with a bandwidth of about 56GHz. among them, S 11 at 358GHz reaches −46.3dB, which exceeds the artificial design level and far exceeds the device design requirements.