.png){kind=icon}
Return
A Multi-Objective Antenna Optimization Method Based on Dual Network-Assisted Differential Evolution with Surface Current-Driven Strategy
.png){kind=icon}
PRE
AI
DOI：10.1109/lawp.2026.3732744.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
A multi-objective optimization method based on dual network-assisted differential evolution (DNADE) is proposed, which aims to enhance the efficiency of mixed-variable antenna optimization. In the proposed DNADE, a novel surface current-driven strategy for physically guided search and two neural networks are incorporated into the differential evolution (DE) algorithm. Specifically, an evolutionary neural network (ENN) is employed as the evolver to learn effective DE search patterns and produce higher-quality candidate solutions. Meanwhile, an online surrogate model based on a multilayer perceptron with a Bayesian last layer (MLP-BLL) serves as the evaluator, thereby reducing the number of expensive electromagnetic (EM) simulations. The effectiveness of DNADE is demonstrated by designing a high-performance dipole antenna with an isolation exceeding 30 dB, a realized gain higher than 8 dBi and a gain flatness less than 1 dB over the operating band of 2.10–3.85 GHz. Compared with other optimization methods, DNADE can achieve the target design with fewer EM simulations.
Keywords:
Multi-objective
antenna optimization
neural network
differential evolution (DE)
surrogate model
Journal
IF:
4.8
Papers:
1.0W
Citations:
2.8W
Organization
Cited Papers
No cited papers available