金属热处理 ›› 2025, Vol. 50 ›› Issue (1): 266-271.DOI: 10.13251/j.issn.0254-6051.2025.01.041

• 计算机应用 • 上一篇    下一篇

机器学习预测退火温度和时间对SUS321不锈钢力学性能的影响

王欢欢1,2, 卢苏君1,2, 李渊3, 徐宁1,2, 朱廷贤3, 王旭1,2, 魏宁4, 吴大莉1,2, 彭威中1,2   

  1. 1.镍钴共伴生资源开发与综合利用全国重点实验室, 甘肃 金昌 737100;
    2.金川镍钴研究设计院有限责任公司, 甘肃 金昌 737100;
    3.金川集团镍合金有限公司, 甘肃 金昌 737100;
    4.金川集团信息与自动化工程有限公司, 甘肃 金昌 737100
  • 收稿日期:2024-07-29 修回日期:2024-11-12 出版日期:2025-01-25 发布日期:2025-03-12
  • 作者简介:王欢欢(1998—),女,工程师,硕士,主要研究方向为金属冶炼,E-mail:annwanghuanhuan@163.com

Machine learning to predict effect of annealing temperature and time on mechanical properties of SUS321 stainless steel

Wang Huanhuan1,2, Lu Sujun1,2, Li Yuan3, Xu Ning1,2, Zhu Tingxian3, Wang Xu1,2, Wei Ning4, Wu Dali1,2, Peng Weizhong1,2   

  1. 1. National Key Laboratory of Ni & Co Associated Minerals Resources Development and Comprehensive Utilization, Jinchang Gansu 737100, China;
    2. Jinchuan Nickel & Cobalt Research and Engineering Institute, Jinchang Gansu 737100, China;
    3. Jinchuan Group Nickel Alloy Co., Ltd., Jinchang Gansu 737100, China;
    4. Jinchuan Group Information and Automation Engineering Co., Ltd., Jinchang Gansu 737100, China
  • Received:2024-07-29 Revised:2024-11-12 Online:2025-01-25 Published:2025-03-12

摘要: 为揭示SUS321不锈钢的力学性能与其热处理工艺之间的复杂关系,基于随机森林模型,通过机器学习方法建立了退火温度、退火时间与SUS321不锈钢力学性能的预测模型。结果表明,模型对SUS321不锈钢抗拉强度、屈服强度和伸长率的预测性能指标R2均超过0.8,表现出较好的预测效果。基于部分依赖图和个体条件期望图的分析表明,随退火温度和退火时间的增加,SUS321不锈钢的抗拉强度和屈服强度呈现下降趋势,而伸长率呈现上升趋势,且退火温度为主要特征参数。

关键词: SUS321不锈钢, 退火处理, 力学性能, 机器学习, 模型解释

Abstract: In order to reveal the complex relationship between mechanical properties and heat treatment process of SUS321 stainless steel, a prediction model between annealing temperature, annealing time and mechanical properties of the SUS321 stainless steel was established by machine learning method based on random forest model. The results show that the prediction validity parameter R2 of the model for the tensile strength, yield strength and elongation of the SUS321 stainless steel exceeds 0.8, showing a good prediction effect. The analysis based on partial dependence plots and individual conditional expectation plots shows that with the increase of annealing temperature and annealing time, the tensile strength and yield strength of the SUS321 stainless steel decrease, while the elongation increases, and the annealing temperature is the main feature parameter.

Key words: SUS321 stainless steel, annealing treatment, mechanical properties, machine learning, model interpretation

中图分类号: