Surrogate-Based EM Optimization Using Neural Networks for Microwave Filter Design
Surrogate-Based EM Optimization Using Neural Networks for Microwave Filter Design
复制标题
DOI：
10.1587/transele.2022mmi0005
复制
发表时间：
2022
期刊：
影响因子：
--
通讯作者：
M. Ohira;Zhewang Ma
中科院分区：
文献类型：
--
作者：
M. Ohira;Zhewang Ma
摘要 提出了一种使用神经网络（NN）的基于代理的电磁（EM）优化，用于计算高效的微波带通滤波器（BPF）设计。本文首先描述了正向问题（EM 分析）和逆向问题（EM 设计），以及 BPF 设计中的两个基本问题。第一个问题是电磁分析是一项耗时的任务，第二个问题是电磁设计高度依赖于电磁分析帮助下进行的结构优化。为了加速优化设计，这里引入了用神经网络构建的正向和逆向模型两种代理模型。因此，只需将合成的耦合矩阵元素输入神经网络，逆模型就可以立即高精度地猜测初始结构参数。然后，正演模型与优化算法相结合，使设计人员能够从初始参数中快速找到最佳结构参数。通过具有多重耦合的典型五阶微带带通滤波器的结构设计验证了基于替代的 EM 优化的有效性。
SUMMARY A surrogate-based electromagnetic (EM) optimization us- ing neural networks (NNs) is presented for computationally e ffi cient microwave bandpass filter (BPF) design. This paper first describes the for- ward problem (EM analysis) and the inverse problems (EM design), and the two fundamental issues in BPF designs. The first issue is that the EM analysis is a time-consuming task, and the second one is that EM design highly depends on the structural optimization performed with the help of EM analysis. To accelerate the optimization design, two surrogate models of forward and inverse models are introduced here, which are built with the NNs. As a result, the inverse model can instantaneously guess initial structural parameters with high accuracy by simply inputting synthesized coupling-matrix elements into the NN. Then, the forward model in conjunction with optimization algorithm enables designers to rapidly find op- timal structural parameters from the initial ones. The e ff ectiveness of the surrogate-based EM optimization is verified through the structural designs of a typical fifth-order microstrip BPF with multiple couplings.