基于最优解区间预筛选的代理模型辅助天线设计优化算法

A Proxy Model Assisted Antenna Optimization Algorithm Based on Optimal Solution Interval Pre-screening

中文关键词:  天线优化  机器学习  差分进化算法  代理模型  最优解区间预筛选

英文关键词:antenna optimization  machine learning  differential evolution algorithm  proxy model  optimal solution interval pre-screening

中文摘要:

      针对天线优化中全波仿真计算耗时过多的问题,文中提出了一种基于数据约束的代理模型辅助进化算法 (SAADC)以实现天线优化设计中效率的提升。首先采用增强随机型差分进化算法,以保证数据生成的随机性与多样性。 进一步通过高斯代理模型对仿真结果进行预测,并利用最优解区间预筛选方法舍弃预测结果中较差的个体,以实现算法 收敛速度的提升。最终利用所提出的SAADC,对三个不同拓扑结构E型贴片天线的带宽与增益进行了优化设计与结果 分析。结果表明,所提出的算法比现有的代理模型优化算法具有更快的优化速度与更佳的优化结果,可满足天线结构高 效优化的实际需求。

英文摘要:

      In this paper, to deal with the problem of time-consuming electromagnetic simulation in antenna optimization, a surrogate model assisted evolution algorithm under data constraints (SAADC) is proposed to improve the efficiency of antenna optimization design. Firstly, the enhanced random differential evolution algorithm is adopted to ensure the randomness and diversity of data generation. Furthermore, the Gaussian proxy model is used to predict the simulation results, and the optimal solution interval pre-screening method is used to discard the poor individuals in the prediction results, so as to improve the convergence speed of the algorithm. Finally, the proposed SAADC is used to optimize the bandwidth and gain of three E-type patch antennas with different topologies. The results show that the proposed algorithm has faster optimization speed and better optimization results than the existing proxy model optimization algorithm, and can meet the practical needs of efficient optimization of antenna structure.