基于机器学习的快淬NdFeB磁体永磁性能分析与预测  

Machine Learning-based Analysis and Prediction of Hard Magnetic Performance for Melt-spun NdFeB Magnets

作　　者：温晋太 胡怀谷 安江山 韩婷 李欣俞 胡季帆 WEN Jintai;HU Huaigu;AN Jiangshan;HAN Ting;LI Xinyu;HU Jifan(School of Materials Science and Engineering,Taiyuan University of Science and Technology,Taiyuan 030024,China;College of Mechanical and Electrical Engineering,Zhengzhou University of Light Industry,Zhengzhou 450002,China;School of Physics,Shandong University,Jinan 250100,China)

机构地区：[1]太原科技大学材料科学与工程学院,太原030024 [2]郑州轻工业大学机电工程学院,郑州450002 [3]山东大学物理学院,济南250100

出　　处：《材料导报》2025年第8期146-152,共7页Materials Reports

基　　金：山西省科技重大专项(202101050201006);山西省重点研发计划(202202050201020)。

摘　　要：NdFeB磁体的永磁性能受合金成分以及工艺参数影响大,机器学习基于数学和信息科学方法,使用现有快淬NdFeB磁体数据来预测NdFeB磁体的磁性能。在本工作中,利用eXtreme Gradient Boosting(XGBoost)算法对快淬NdFeB磁体永磁性能进行分析。结果表明,相较于其他机器学习模型,利用集成学习XGBoost算法开发出的机器学习模型对快淬NdFeB磁体永磁性能的预测结果精度更高,稳定性更好。同时还利用该XGBoost模型,优化预测出新的具有较高永磁性能的快淬NdFeB材料。The permanent magnetic performance of NdFeB magnets strongly depends on the alloy composition and process parameters.Machine lear-ning is based on mathematical and information science methods,using existing rapidly quenched NdFeB magnet data to predict the magnetic properties of NdFeB magnets.In this work,the eXtreme Gradient Boosting(XGBoost)algorithm was used to analyze the permanent magnet performance of rapidly quenched NdFeB magnets.The results show that the machine learning model developed from ensemble learning XGBoost algorithm has a higher prediction accuracy and better stability for the permanent magnet performance of rapidly quenched NdFeB magnets than other algorithms.Then the machine model based on XGBoost algorithm proposed here is utilized to optimize and predict new melt-spun NdFeB mate-rials with high permanent magnetic performance.

关 键 词：钕铁硼 磁性能 合金成分 工艺参数 极端梯度提升 

分 类 号：TM273[一般工业技术—材料科学与工程]