Power electronic converters are key components in modern power systems. Due to the special characteristics such as processing power flow and operating in switching mode, the layout and routing of power electronic converters are still mainly performed manually, with a low level of automation. This paper reviews the research related to the layout and routing of the main circuits of power electronic converters, distills the core requirements, categorizes the research into traditional layout and deep learning layout methods, and analyzes the characteristics of different approaches. In addition, the paper also proposes a hierarchical layout and routing design framework that can effectively reduce the design parameter space and fully leverage the advantages of design tools such as expert knowledge, deep reinforcement learning, and traditional algorithms. Based on this algorithmic framework, the paper demonstrates the design cases of the self-developed software “Ai BuDao” in the scenario of low-power circuits, preliminarily showing the effectiveness of this design framework.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Intelligent Layout and Routing of Power Electronic Converters: A Technical Review and Cutting-Edge Exploration

  • Yi Shang,
  • Yu Chen,
  • Hanwen Chen,
  • Zhisen Zhu,
  • Yong Kang

摘要

Power electronic converters are key components in modern power systems. Due to the special characteristics such as processing power flow and operating in switching mode, the layout and routing of power electronic converters are still mainly performed manually, with a low level of automation. This paper reviews the research related to the layout and routing of the main circuits of power electronic converters, distills the core requirements, categorizes the research into traditional layout and deep learning layout methods, and analyzes the characteristics of different approaches. In addition, the paper also proposes a hierarchical layout and routing design framework that can effectively reduce the design parameter space and fully leverage the advantages of design tools such as expert knowledge, deep reinforcement learning, and traditional algorithms. Based on this algorithmic framework, the paper demonstrates the design cases of the self-developed software “Ai BuDao” in the scenario of low-power circuits, preliminarily showing the effectiveness of this design framework.