基于人工神经网络的热斗篷设计  

Design of Thermal Cloaks Based on Artificial Neural Network

作　　者：单庆茹 丁宇 王鹤文 周欣智 许云宇 王军[1] 夏国栋[1] SHAN Qingru;DING Yu;WANG Hewen;ZHOU Xinzhi;XU Yunyu;WANG Jun;XIA Guodong(MOE Key Laboratory of Enhanced Heat Transfer and Energy Conservation,Beijing Key Laboratory of Heat Transfer and Energy Conversion,Beijing University of Technology,Beijing 100124,China)

机构地区：[1]北京工业大学,传热强化与过程节能教育部重点实验室暨传热与能源利用北京市重点实验室,北京100124

出　　处：《工程热物理学报》2025年第3期982-987,共6页Journal of Engineering Thermophysics

摘　　要：基于体材料的双层热隐身斗篷由大块各向同性材料构成,可以避免变换热学的复杂计算,但是由于内层材料热导率不完全绝热,在背景材料热导率较低时,其热隐身效果较差。本文基于机器学习方法,构建关于热隐身斗篷参数与隐身效果之间的人工神经网络,智能地学习外层材料的热导率以及斗篷尺寸与双层斗篷热隐身效果之间的关系并输出功能函数,计算得到能实现较好热隐身效果的外层材料热导率或斗篷尺寸大小,并模拟验证了这一机器学习结果的准确性,实现了双层斗篷热隐身效果的优化。A bilayer thermal cloak consists of an inner layer with very low thermal conductivity and an outer layer with higher thermal conductivity,which can eliminate the external-field distor-tion.However,the thermal cloak phenomenon could not be achieved for lower surrounding thermal conductivity in the bilayer thermal cloak,because the inner layer could not be completely insulated.In this paper,based on the method of machine learning,the performance of the bilayer thermal cloak has been optimized.An artificial neural network is established to intelligently learn the relation be-tween each layer’s thermal conductivity and the cloaking performances.The optimized parameters for a better performance of the bilayer thermal cloak can be obtained,which is verified by numerical simulations.

关 键 词：热斗篷 人工神经网络 热导率 模拟 

分 类 号：TK124[动力工程及工程热物理—工程热物理]