Trust-Region-based Multi-Branch Machine Learning-Assisted Optimization for Ultra-Wideband Antenna

In this paper, a trust-region-based multi-branch machine learning-assisted optimization (TRMB-MLAO) method is proposed for ultra-wideband antenna design. By introducing techniques including trust-region design space constraint strategy, multi-branch lower confidence bound acquisition function and multi-point acquisition function composition into machine learning-assisted optimization scheme, the proposed method achieves better convergence speed and antenna performance compared with conventional algorithm. An ultra-wideband antenna is introduced to validate the effectiveness of the proposed method.