正在加载图片...
《工程科学学报》录用稿,htps:/doi.org/10.13374/i,issn2095-9389.2021.06.22.001©北京科技大学2020 工程科学学报DO: 基于机器学习的产品质量在线智能监控方法 徐钢2),黎敏),吕志民),徐金梧)@ 1)北京科技大学钢铁共性技术协同创新中心,北京100083 2)苏州宝联重工有限公司,江苏苏州215131 ☒通信作者,E-mail:jwxu@ustb.edu.cm 摘要目前,我国钢铁企业主要采用“事后”抽检方式,对产品质量进行出厂前检骏。由卡无法对所有产品实现质量 检验,常发生客户索赔和退货,导致钢铁企业重大经济损失。为了提高产品质量的稳定性和阿靠性,利用机器学习方 法实现产品质量在线监控、在线优化和在线预设定,是钢铁企业目前亟待解决的关键最术。针对企业需求,提出基于软 超球体算法的产品质量异常在线识别和异常原因诊断方法、基于流形学习的工艺参数在线优化方法和基于多变量统计 过程控制的工艺规范制定方法。通过将上述方法进行系统集成,并利用工业互联网技术和大数据分析方法,研发了产 品质量在线智能监控系统。目前该系统己在钢铁企业十余条生产线上维广应用,质量在线判定的准确率达到99.2%, 在线检测时间不到0.1秒。 关键词大数据分析:机器学习:质量在线监控:质量在线优? 质量设计 分类号TP274 On-line Intelligent Monitoring Method of Product Quality based on Machine Learning XU Gang'2,LI Min,LV Zhimin 1) Collaborative Innovation Center of echnology,University of Science and Technology Beijing,Beijing 100083,China Suzhou Baolian Heavy Indus 15131,China Corresponding author, .edu.cn ABSTRACT In recent years,Chinese iron and steel enterprises mainly adopt "sampling after the event"method to inspect the product quality before leaving the factory.Because of the lack to achieve quality inspection for all products,customers often claim and return the defective products,which leads to the major economic losses of steel enterprises.In order to improve the stability and reliability of product quality,using machine learning methods to realize on-line monitoring,on-line optimization and on-line preset of product quality are the key technology to be solved in iron and steel enterprises.Therefore,on-line identification and on-line diagnosis of abnormal product quality based on soft hypersphere,on-line optimization of process parameters based on manifold learning,and process specification formulation based on multivariate statistical process control are proposed. 收篇日期: 盖演自:国家“863"重点课题:面向钢铁行业的全流程智能分析技术研究(2013AA040704),自然科学基金: (51004013)、(51204018)工程科学学报 DOI: 基于机器学习的产品质量在线智能监控方法1 徐 钢 1,2),黎 敏 1) ,吕志民 1),徐金梧 1)  1) 北京科技大学钢铁共性技术协同创新中心,北京 100083 2) 苏州宝联重工有限公司, 江苏苏州 215131  通信作者,E-mail: jwxu@ustb.edu.cn 摘 要 目前,我国钢铁企业主要采用“事后”抽检方式,对产品质量进行出厂前检验。由于无法对所有产品实现质量 检验,常发生客户索赔和退货,导致钢铁企业重大经济损失。为了提高产品质量的稳定性和可靠性,利用机器学习方 法实现产品质量在线监控、在线优化和在线预设定,是钢铁企业目前亟待解决的关键技术。针对企业需求,提出基于软 超球体算法的产品质量异常在线识别和异常原因诊断方法、基于流形学习的工艺参数在线优化方法和基于多变量统计 过程控制的工艺规范制定方法。通过将上述方法进行系统集成,并利用工业互联网技术和大数据分析方法,研发了产 品质量在线智能监控系统。目前该系统已在钢铁企业十余条生产线上推广应用,质量在线判定的准确率达到 99.2%, 在线检测时间不到 0.1 秒。 关键词 大数据分析;机器学习;质量在线监控;质量在线优化;产品质量设计 分类号 TP274 On-line Intelligent Monitoring Method of Product Quality based on Machine Learning XU Gang1,2) , LI Min1) , LV Zhimin1),XU Jinwu1)  1) Collaborative Innovation Center of Steel Technology, University of Science and Technology Beijing, Beijing 100083, China 2) Suzhou Baolian Heavy Industry Co. Suzhou, 215131, China  Corresponding author, E-mail: jwxu@ustb.edu.cn ABSTRACT In recent years, Chinese iron and steel enterprises mainly adopt "sampling after the event" method to inspect the product quality before leaving the factory. Because of the lack to achieve quality inspection for all products, customers often claim and return the defective products, which leads to the major economic losses of steel enterprises. In order to improve the stability and reliability of product quality, using machine learning methods to realize on-line monitoring, on-line optimization and on-line preset of product quality are the key technology to be solved in iron and steel enterprises. Therefore, on-line identification and on-line diagnosis of abnormal product quality based on soft hypersphere, on-line optimization of process parameters based on manifold learning, and process specification formulation based on multivariate statistical process control are proposed. 1收稿日期: 基金项目: 国家“863”重点课题:面向钢铁行业的全流程智能分析技术研究(2013AA040704), 自然科学基金: (51004013)、(51204018) 《工程科学学报》录用稿,https://doi.org/10.13374/j.issn2095-9389.2021.06.22.001 ©北京科技大学 2020 录用稿件,非最终出版稿
向下翻页>>
©2008-现在 cucdc.com 高等教育资讯网 版权所有