A Fast Design and Optimization Method Based on Surrogate Model and Machine Learning

被引：1

作者：

Li, Wen Xi ^{[1
]}

Li, Ying ^{[1
]}

Yan, Ran ^{[1
]}

Luo, Yong ^{[1
]}

机构：

[1] Univ Elect Sci & Technol China, Sch Elect Sci & Engn, Chengdu, Peoples R China

来源：

IVEC 2021: 2021 22ND INTERNATIONAL VACUUM ELECTRONICS CONFERENCE | 2021年

关键词：

surrogate model; machine learning; genetic algorithm; NSGA-II;

D O I：

10.1109/IVEC51707.2021.9722392

中图分类号：

TM [电工技术]; TN [电子技术、通信技术];

学科分类号：

0808 ; 0809 ;

摘要：

Design and optimize an electromagnetic device usually relies on the use of simulation software and intelligent optimization algorithm. Designer need to create a model of device, assign relevant design parameters, then the computer will instead of designer to simulate device and get the optimal result. But this method has some obvious disadvantages. First, how to create an appropriate model relies on designers' experience. Second, it costs a lot of time to do simulation calculations. Therefore, this method has long design cycles and low design efficiency. In order to overcome these weaknesses, this paper introduce a faster design method based on surrogate model and machine learning. Compared with the old method, it can be obviously found that the new method has lower computational complexity and higher convergence speed.

引用

页数：2