NEC和非堵塞式主从并行遗传算法应用于 天线自动设计的研究

利用优化算法和天线数值计算方法实现对天线结构的自动设计(Automated design)是现代天线研究的一个重要趋势.本文讨论了天线自动设计的原理和流程,采用遗传算法(Genetic algorithms)和NEC(Numerical electromagnetics code)天线数值计算程序,建立了一套天线自动设计软件平台.采用并行计算技术提高自动设计效率,搭建了一套Beowulf并行计算机系统,首次提出非堵塞式主从并行遗传算法的实现方案.以对锥削螺旋－圆锥喇叭天线的自动设计为例,结果表明该自动设计软件平台具备对复杂天线进行准确和有效设计的能力.16节点的并行效率达到了82.25%,超过同类研究结果.

A Study of Applying NEC and Non-Blocking Master-Slave Parallel Genetic Algorithms to Automated Antenna Design

Automated antenna design based on optimization engines and antenna modeling programs has already become the latest trend of antenna design.An antenna automated design software package using Genetic Algorithms coupled with NEC (Numerical electromagnetics code) was introduced.To elevate efficiency of antenna design,a Beowulf parallel computing system was built,and a new proposal,namely non-blocking master-slave parallel Genetic Algorithm,was explored.The principle and methods of automated antenna design based on Parallel Genetic Algorithms was discussed.An example of automatically designing a tapered helix-conical horn antenna by the package was presented.Results show the package has ability to accurately and efficiently design complex antennas,and its parallel efficiency could be maintained at about 82.25% even when up to 16 processors were used,which is higher than corresponding methods.