Considering the defects of particle swarm optimization (PSO) and multi-objective evolutionary algorithm (EA) based on decomposition (MOEA/D), this article proposes an improved multi-objective EA MOEA/D-GPSO for compact log-periodic dipole array (LPDA) design. MOEA/D-GPSO decomposes multiple objectives into a number of single-objective optimization problems. Each particle deals with one sub-problem. All the particles are divided into a few groups, and each particle has several neighboring particles. During the search, a new solution is constructed by learning information from the random non-dominated solutions found by its own neighbors and group. ZDT instances have been introduced to verify the effectiveness of MOEA/D-GPSO with respect to other outstanding multi-objective EAs. Further, two miniaturized LPDA designs for the application of digital video broadcasting-terrestrial (DVB-T) (470–790 MHz), a tri-objective compact LPDA, and a novel LTE800-refused compact LPDA, respectively, are presented, showing their good performance over other similar designs and promising prospect of the proposed algorithm for high-dimensional and multi-functional LPDA design.

中文翻译：

---

使用 MOEA/D-GPSO 的三目标紧凑对数周期偶极子阵列天线设计

针对粒子群优化(PSO)和基于分解的多目标进化算法(EA)(MOEA/D)的缺陷，本文提出了一种改进的多目标EA MOEA/D-GPSO紧凑对数周期偶极阵列( LPDA) 设计。MOEA/D-GPSO 将多个目标分解为多个单目标优化问题。每个粒子处理一个子问题。所有的粒子被分成几组，每个粒子有几个相邻的粒子。在搜索过程中，通过从其自己的邻居和组找到的随机非支配解决方案中学习信息来构建新的解决方案。引入了 ZDT 实例来验证 MOEA/D-GPSO 相对于其他出色的多目标 EA 的有效性。更多，