A Batched Antenna Topology Optimization Method Based on Multi-Weighted Acquisition Function

A machine learning-assisted antenna topology optimization algorithm based on multi-weighted lower confidence bound acquisition function (MW-MLAO) is proposed. The multiweighted acquisition function is introduced to solve constrained design tasks efficiently. The algorithm is then introduced to design neutral line decoupling topology of an ultra wideband antenna array. The isolation of the antenna array has been improved for at least 8 dB, and the lowest $\left|S_{21}\right|$ reaches below at least $\mathbf{- 1 8} \mathbf{~ d B}$ between 3.6 and 9.8 GHz.