人工智能在运算放大器设计中的应用研究  

Application of Artificial Intelligence in Operational Amplifier Design

作　　者：乔寓然 王鹏[2] 谢家志[2] QIAO Yuran;WANG Peng;XIE Jiazhi(School of Computer Science,National University of Defense Technology,Changsha 410073,P.R.China;The 24th Research Institute of China Electronics Technology Group Corporation,Chongqing 400060,P.R.China)

机构地区：[1]中国人民解放军国防科技大学计算机学院,长沙410073 [2]中国电子科技集团公司第二十四研究所,重庆400060

出　　处：《微电子学》2024年第4期653-658,共6页Microelectronics

基　　金：国家自然科学基金资助项目(60906009);中国博士后科学基金资助项目(20090451423);重庆科委基金资助项目(CSTC2010AA2004)。

摘　　要：近年来,人工智能技术在集成运算放大器设计中的应用成效逐步显现。本文阐述了利用神经网络技术,辅助运算放大器设计的基本原理和研究现状。重点描述了如何通过使用深度Q网络结合粒子群优化、多目标遗传算法结合SPICE优化、贝叶斯优化等人工智能设计方法,提高运算放大器设计效率、获得更好的器件性能,并对人工智能技术对芯片设计未来发展的影响进行了展望。可为采用人工智能技术辅助放大器设计的模拟集成电路设计人员提供参考。The application of artificial intelligence in integrated operational amplifier design has been extensively investigated in the recent years.Herein,the background,basic principles,and progress made in operational amplifier design using neural networks are elucidated.Furthermore,this study demonstrates that combining deep Q-networks and particle swarm optimization,integrating multi-objective genetic algorithm and SPICE optimization,and utilizing Bayesian optimization and other artificial intelligence methods can enhance both the performance and design efficiency of operational amplifiers.In addition,the development of chip design trends based on artificial intelligence is discussed.This study provides analog integrated circuit designers with a reference for circuit design based on artificial intelligence.

关 键 词：运算放大器 模拟集成电路 人工智能 神经网络 贝叶斯优化 

分 类 号：TN432[电子电信—微电子学与固体电子学]