机器学习辅助辐射特性定向调控优化设计研究与应用综述  

A Review on the Research and Application of Machine Learning Assisted Optimal Design for Directional Regulation of Radiation Properties

作　　者：杜佳欣 王富强[1] 张鑫平 宋锦涛 DU Jiaxin;WANG Fuqiang;ZHANG Xinping;SONG Jintao(School of Energy Science and Engineering,Harbin Institute of Technology,Harbin Hei Longjiang 150001)

机构地区：[1]哈尔滨工业大学能源科学与工程学院,黑龙江哈尔滨150001

出　　处：《东北电力大学学报》2024年第6期63-73,共11页Journal of Northeast Electric Power University

基　　金：国家自然科学基金面上项目(52476067);国家自然科学基金面上项目(52076064)。

摘　　要：实现多波段辐射特性的高效精准调控是军事伪装、航空航天、太阳能等领域的共性科学难题。传统辐射特性调控通过低效试错的方式优化官能团或微纳结构,费时费力且难以获得最佳辐射特性。机器学习的出现颠覆传统的优化方法,以模拟大脑学习思考的方式,极大程度地提高辐射特性优化设计的效率。首先,讨论辐射特性调控中的机器学习算法,评价其在准确性、扩展性和效率等方面的优势与挑战;然后,系统地总结机器学习与辐射特性定向调控融合应用的先进成果,包括前向辐射响应预测和材料定向优化设计;最后,探讨辐射特性调控与机器学习结合的研究热点和未来发展方向。通过现有文献,为辐射特性定向调控与机器学习算法的设计和应用提供参考,为辐射特性定向调控进一步优化与创新提出建议。How to achieve efficient and precise control of multi-band radiation properties is a common scientific challenge in military camouflage,aerospace,solar energy and other fields.Conventional radiation property control often uses inefficient trial-and-error optimisation of functional groups or micro-nanostructures,which is time-consuming,laborious and difficult to obtain the best radiation properties.The emergence of machine learning has overturned the traditional optimisation methods and greatly improved the efficiency of radiation property optimisation and design by simulating the brain's learning and thinking.In this paper,machine learning algorithms in radiation property regulation are discussed in detail,and their advantages and challenges in terms of accuracy,scalability and efficiency are evaluated;the advanced results of the fusion of machine learning and radiation property directional regulation are summarised in a systematic way,including forward radiation response prediction and material directional optimal design;and finally,the hot spots of the research on the combination of radiation property regulation and machine learning and the direction of future development are explored.By reviewing the existing literature,this paper provides a reference for the design and application of radiation property directional regulation and machine learning algorithms,and makes suggestions for further optimisation and innovation of radiation property directional regulation.

关 键 词：机器学习 辐射特性 定向设计 数据驱动 优化算法 

分 类 号：TB303[一般工业技术—材料科学与工程]