|
| 基于机器学习的锻件热加工质量预测研究现状与发展趋势 |
| Research Status and Development Trend of Forging Quality Prediction in Hot Working Process Based on Machine Learning |
| Received:June 15, 2024 |
| DOI：10.3969/j.issn.1674-6457.2025.02.022 |
| 中文关键词: 热加工 锻件 质量预测 机器学习 数据机理 数字孪生 大模型 |
| 英文关键词: hot working forging quality prediction machine learning data mechanism digital twin large scale model |
| 基金项目:国家重点研发计划(2023YFB3307600);国家自然科学基金面上项目(52075400);湖北隆中实验室自主创新项目(2022ZZ-04);国家111计划项目课题(B17034);湖北省重点研发计划(2021BAA200);湖北省技术创新重大专项(2022AAA001);湖北省科技创新人才及服务专项(2022EJD012) |
| Author Name | Affiliation | | LIU Hongtao | Hubei Longzhong Laboratory, Hubei Xiangyang 441022, China
a.Hubei Key Laboratory of Modern Automotive Parts Technology, b.Hubei Engineering Center of Materials Green Forming Technology and Equipment, Wuhan University of Technology, Wuhan 430070, China | | ZHANG Rui | China National Erzhong Group Deyang Wanhang Die Forging Co., Ltd., Sichuan Deyang 618000, China | | HU Zhili | Hubei Longzhong Laboratory, Hubei Xiangyang 441022, China
a.Hubei Key Laboratory of Modern Automotive Parts Technology, b.Hubei Engineering Center of Materials Green Forming Technology and Equipment, Wuhan University of Technology, Wuhan 430070, China |
|
| Hits: 16 |
| Download times: 3 |
| 中文摘要: |
| 锻件热加工产品结构复杂，性能要求高，工艺窗口窄，成形难度高，导致产品质量波动大。将机器学习技术应用于锻件热加工质量预测是控制产品质量、提高产品良品率的有效解决手段。综述了国内外最近几年关于机器学习的锻件热加工质量预测研究进展。首先阐述了锻件热加工质量定义，其次列举了应用于锻件热加工质量预测的常见方法。重点论述了机器学习技术在锻件热加工工艺参数预测及组织性能预测2个方面的研究现状和进展，重点关注样本容量、网络拓扑结构、算法特点对预测结果的影响，及相应解决措施。随着新一代人工智能技术的发展，探讨了基于机器学习的数据-机理融合驱动质量预测方法、数字孪生驱动质量预测方法以及工业大模型驱动质量预测方法等3个方面近期的研究进展，分析了工业大模型驱动质量预测方法在未来可能面对的机遇和挑战。 |
| 英文摘要: |
| The complex structure, high performance requirements, narrow process window and high forming difficulty of forging products during hot working lead to large fluctuations in product quality. Applying machine learning technology to quality prediction of hot working forgings is an effective means to control product quality and improve product yield. Therefore, the work aims to review the research progress of machine learning on quality prediction of hot working forgings in recent years. Firstly, the definition of hot working quality of forgings was described, and then the common methods applied to hot working quality prediction of forgings were listed. The research status and progress of machine learning technology in hot working process parameter prediction and microstructure performance prediction of forgings were mainly discussed, focusing on the impact of sample size, network topology and algorithm characteristics on the prediction results, and the corresponding solutions. With the development of the new generation of artificial intelligence technology, the recent research progress of machine learning-based data-mechanism fusion driven quality prediction method, digital twin driven quality prediction method and industrial large model driven quality prediction method were also discussed. The opportunities and challenges that industrial large model driven quality prediction method may face in the future were analyzed. |
|
View Full Text
View/Add Comment Download reader |
| Close |
|
|
|
渝公网安备 50010702501719号