计算机集成制造系统 ›› 2018, Vol. 24 ›› Issue (第7): 1850-1857.DOI: 10.13196/j.cims.2018.07.027

• 当期目次 • 上一篇    下一篇

基于多核学习的装配工艺过程重用

王裴岩1,2,张桂平1,翟顺龙3,蔡东风1,2   

  1. 1.沈阳航空航天大学人机智能研究中心
    2.南京航空航天大学计算机科学与技术学院
    3.沈阳格微软件有限责任公司知识工程及服务事业部
  • 出版日期:2018-07-31 发布日期:2018-07-31
  • 基金资助:
    辽宁省自然科学基金资助项目(20170540705)。

Multiple kernel learning based assembly process reuse

  • Online:2018-07-31 Published:2018-07-31
  • Supported by:
    Project supported by the Natural Science Foundation of Liaoning Province,China(No.20170540705).

摘要: 为了最大程度复用历史工艺过程,提高工艺设计效率,对装配工艺过程重用问题进行了研究,提出了一种基于多核学习的方法。该方法利用了工艺规程名称、规程编号、设计人与装配零件表等直接获取特征,不需要人工确定特征的表达符号集与标注数据;通过定义多个核函数,从不同视角衡量工艺过程的可重用性,并利用工艺大纲文件间的重用度作为指导信息,优化多核组合参数。在47 828份飞机装配工艺规程数据上,多核学习方法能够有效地对工艺规程文件进行筛选与排序,能够保证排序靠前的结果具有较高的重用度,Top1重用度可达0.3811。实验结果证明了规程文件命名规律、企业工艺文件管理规则、装配零件表等信息在工艺过程复用中的有效性。

关键词: 装配工艺过程重用, 核函数, 多核学习, 飞机装配

Abstract: To maximize the historical process reuse and improve the efficiency of process design,the assembly process reuse problem was researched,and a multiple kernel learning method was proposed.In this method,the features were obtained directly with kernel functions included process name,process number,process designer and assembly parts list.Through defining multiple kernel functions,the reusability of assembly process were measured with different perspective,and the multiple kernel combination parameters were optimized by using process outline file for guidance.Experimental results on 47 828 aircraft assembly processes showed that the multiple kernel learning method could effectively select and sort the process files,and could ensure the results of higher ranking with higher reusability.The reuse ratio of Top1 result could reach 0.3811.The effectiveness of information such as the process naming and cognition rule,the enterprise process document management rule and the design task assignment principle in the process reuse of assembly process was proved.

Key words: assembly process reuse, kernel function, multiple kernel learning, aincraft assembly

中图分类号: