面向热力性能调控的航天器功能梯度结构智能设计研究  

Intelligent design of spacecraft functionally gradient structures for thermal-mechanical performance regulation

作　　者：刘海洲 赵阳[1] 黄意新[1] LIU HaiZhou;ZHAO Yang;HUANG YiXin(School of Astronautics,Harbin Institute of Technology,Harbin 150001,China)

机构地区：[1]哈尔滨工业大学航天学院,哈尔滨150001

出　　处：《中国科学:物理学、力学、天文学》2025年第2期146-158,共13页Scientia Sinica Physica,Mechanica & Astronomica

基　　金：民用航天技术预先研究计划(编号:D020304);国家自然科学基金(编号:52205257)资助项目。

摘　　要：以高精高稳航天器中含复杂热源的精密载荷安装结构为背景,提出了一种基于物理信息神经网络的功能梯度结构智能设计方法,不依赖热控装置,通过优化材料梯度规律对载荷安装区域局部温度和结构基频进行协同调控.考虑多种热源和载荷分布情况,建立热传导与振动问题控制方程,将之与局部温度优化目标、结构基频要求、材料组分约束一起构建物理信息神经网络损失函数,以材料梯度规律为设计变量,采用物理信息神经网络进行逆问题求解.结果表明,合理设计材料梯度规律可在不依赖外部热控装置的情况下精确控制结构内指定位置局部温度并保证结构基频.本文所提出的方法有望降低航天器结构对热控与减隔振装置需求,提升结构热力稳定性.Against the background of precision payload mounting plates with complex heat sources in high-precision and highstability spacecraft, an intelligent design method for functionally graded structures based on physics-informed neuralnetworks is presented. Independent of thermal control devices, the local temperature in the load installation area and thefundamental frequency of the structure are regulated by optimizing the material gradient distribution. A loss function isestablished that includes the heat-conduction governing equation, local temperature and fundamental frequencyoptimization objectives, and material volume fraction constraints, considering multiple heat source distributions. Theinverse problem with material gradient distribution as a design variable is solved using physics-informed neuralnetworks. The results show that a reasonable design of the material gradient distribution can accurately regulate the localtemperature at specific locations and the fundamental frequency without relying on external thermal control devices. Theproposed method is expected to reduce the need for thermal vibration control devices for spacecraft structures andimprove structural thermal-mechanical stability.

关 键 词：物理信息神经网络 航天器结构 功能梯度材料 热传导 材料优化设计 

分 类 号：V414.1[航空宇航科学与技术—航空宇航推进理论与工程]