一种基于Kriging模型和受限差分进化的电磁结构快速优化算法

发布时间：2018-05-05 14:05

本文选题：电磁结构 + 优化算法　；参考：《电波科学学报》2017年03期

【摘要】：进化算法在各类电磁结构优化设计中有着广泛的应用,但由于需要在参数空间中进行随机搜索并仿真试探,优化效率普遍较低.针对这一问题,提出受限差分进化(Differential Evolution,DE)算法与Kriging代理模型相结合的电磁结构快速优化算法.算法根据参考设计结果建立圆柱管道空间,通过参数变换将进化区域限制在管道内部.Kriging模型学习管道内样本及其仿真数据,代替电磁仿真快速预测进化产生下一代种群的响应.相比整个参数空间,该算法DE寻优和Kriging学习的区域被显著减小,优化效率得到提升.通过一个波导双孔定向耦合器的优化设计,表明该方法的求解质量和收敛速度优于现有算法.
[Abstract]:The evolutionary algorithm is widely used in the optimization design of various electromagnetic structures. However, due to the need for random search and Simulation in the parameter space, the optimization efficiency is generally low. In view of this problem, a fast optimization calculation of the electromagnetic structure combined with the Differential Evolution (DE) algorithm and the Kriging agent model is proposed. The algorithm establishes a cylindrical pipe space based on the reference design results. Through the parameter transformation, the evolutionary region is restricted to the.Kriging model inside the pipeline to learn the sample and its simulation data, instead of the electromagnetic simulation to quickly predict the response of the next generation of the population. Compared with the whole parameter space, the algorithm DE is optimized and the Kriging learning area is found. The optimal design of a waveguide double hole directional coupler shows that the solution quality and convergence speed of the method are better than those of the existing algorithms.

【作者单位】：安徽工程大学电气工程学院通信工程系;
【基金】：安徽省高等教育提升计划项目(TSKJ2014B05,TSKJ2015B19)
【分类号】：TN622;TP18

【相似文献】