loading page

Schematic Turing Test for Cognitive Power Electronics
  • Minjie Chen ,
  • Dak Cheung Cheng
Minjie Chen
Princeton University

Corresponding Author:[email protected]

Author Profile
Dak Cheung Cheng
Princeton University

Abstract

This paper explores schematic Turing test for realizing artificial general intelligence (AGI) of power electronics design. We first propose Schematic Turing Test (STT) as one critical step to examine the competence of high-level generative artificial intelligence for power electronics, then present PowerAGI-a machine learning framework for implementing artificial general intelligence (AGI) in power electronics. The PowerAGI includes two building blocks, PowerVision and SchemaNet. PowerVision is a schematic recognition tool which converts human-readable images into machine-readable netlists through component and wiring recognition. SchemaNet is a schematic and netlist database for rapid topology classification. Once scaled, the PowerVision tool and the SchemaNet database can enable artificial general intelligence models, such as large language models, to learn about power electronics fundamental principles through a mix of information in different format, including text descriptions, schematic recognition, computer simulations, and image illustrations.
31 May 2024Submitted to TechRxiv
07 Jun 2024Published in TechRxiv