机器学习在永磁材料研究中的应用进展  

Application of machine learning in permanent magnetic material

作　　者：李金栋 郝永勤 孙旭 沈鹏 韩瑞[1] 周栋[1] LI Jindong;HAO Yongqin;SUN Xu;SHEN Peng;HAN Rui;ZHOU Dong(Division of Functional Materials,Central Iron&Steel Research Institute,Beijing 100081,China;Beijing Institute of Aerospace Control Devices,Beijing 100039,China;School of Physics,Peking University,Beijing 100871,China)

机构地区：[1]钢铁研究总院有限公司功能材料研究院,北京100081 [2]北京航天控制仪器研究所,北京100854 [3]北京大学物理学院,北京100871

出　　处：《功能材料》2025年第1期1064-1074,共11页Journal of Functional Materials

摘　　要：永磁材料在现代工业与科学技术中发挥了重要的作用。近年来,借助机器学习方法在预测和优化永磁材料的制备与应用方面取得了巨大的发展。较为全面的综述了机器学习在永磁材料研究中的应用,介绍了机器学习的学习流程和常用的机器学习算法,综述了机器学习技术在微观特性分析与结构优化、磁性能预测与成分优化、探索新材料等方面的研究进展。提出了机器学习在永磁材料领域所面临的问题,包括数据维度高、样本量有限、噪音干扰大、缺失值较多等。在未来研究中,应深入研究并探索新的算法和优化策略,扩充数据集规模,以及结合智能化实验技术加快永磁材料的研发与改进。Permanent magnetic materials play an important role in modern industry and technology.In recent years,tremendous progress has been made in predicting and optimizing the preparation and application of permanent magnetic materials using machine learning methods.This paper comprehensively reviews the application of machine learning in research on permanent magnetic materials,introduces the learning process of machine learning and commonly used machine learning algorithms,and summarizes the research progress of machine learning technology in microstructure optimization and characterization analysis,magnetic properties prediction and component optimization,exploring new materials.This paper raises the issues faced by machine learning in the field of permanent magnet materials,including high data dimension,limited sample size,large noise interference,and more missing values.In future research,new algorithms and optimization strategies should be deeply studied and explored,the scale of the dataset should be expanded,and intelligent experimental techniques should be combined to accelerate the research and development and improvement of permanent magnet materials.

关 键 词：永磁材料 机器学习 磁特性分析 磁性能预测 材料设计 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]