• CN:11-2187/TH
  • ISSN:0577-6686

机械工程学报 ›› 2017, Vol. 53 ›› Issue (18): 191-201.doi: 10.3901/JME.2017.18.191

• 交叉与前沿 • 上一篇    下一篇

一类考虑返修和产品多态性的多级串联制造系统建模与性能分析

张恒, 李爱平, 刘雪梅, 徐立云   

  1. 同济大学现代制造技术研究所 上海 200092
  • 收稿日期:2016-07-12 修回日期:2017-03-10 出版日期:2017-09-20 发布日期:2017-09-20
  • 通讯作者: 张恒(通信作者),男,1990年出生,博士研究生.主要研究方向为制造系统与自动化等.E-mail:zhangheng1990daniel@foxmail.com
  • 作者简介:李爱平,女,1951年出生,博士,教授,博士研究生导师.主要研究方向为制造系统与自动化、数字化设计与制造、制造信息技术与工程等.E-mail:limuzi@tongji.edu.cn;刘雪梅,女,1969年出生,博士,教授,博士研究生导师.主要研究方向为数字化设计与制造、制造信息技术与工程、制造系统与自动化等.E-mail:liuxuemei@tongji.edu.cn;徐立云,男,1973年出生,博士,教授,博士研究生导师.主要研究方向为智能制造、系统建模与优化、产品数字化设计与管理等.E-mail:lyxu@tongji.edu.cn
  • 基金资助:
    国家科技重大专项(2011ZX04015-022,2013ZX04012-071)和上海市科委(上海市科技成果转化与产业化)(15111105500)资助项目。

Modeling and Performance Evaluation of a Multistage Serial Manufacturing System Considering Rework and Product Polymorphism

ZHANG Heng, LI Aiping, LIU Xuemei, XU Liyun   

  1. Institute of Advanced Manufacturing Technology, Tongji University, Shanghai 200092
  • Received:2016-07-12 Revised:2017-03-10 Online:2017-09-20 Published:2017-09-20

摘要: 针对一类考虑返修和产品多态性的多级串联制造系统性能分析问题,在考虑系统设备故障基础上,重点研究产品的返修和报废以及产品的多态性导致工位正常运行时加工状态差异对系统性能的影响,提出一种基于扩展Petri网(Extended-stochastic Petri nets,ESPN)的系统模块化建模方法,解决传统利用Markov模型和排队网络模型建模难以清晰描述系统内部产品流转换导致复杂状态变化的问题。将ESPN模型同构于嵌入Markov链得到工位状态空间,推导产品多态性导致的工位不同加工状态相对激发概率分布。以双工位单缓冲区的两级串联制造系统单元模型性能分析为基础,考虑缓冲区两边系统对单元模型的影响,基于重叠分解法对系统模型进行迭代求解,分析系统各项性能指标。某企业汽车动力总成装配线的实例验证了该建模与分析方法的有效性。

关键词: 产品多态性, 多级串联制造系统, 返修, 扩展随机 Petri 网, 性能分析, 重叠分解法

Abstract: Modeling and performance evaluation of a multistage serial manufacturing system with consideration of rework and products polymorphism is focused. In addition to considering equipment failure, the impacts of products rework and scrap, processing differences resulting from products polymorphism are emphasized. To solve the problem that the system model cannot accurately describe the complex state change resulting from the conversion of internal product flows, which is formulated by traditional methods based on Markov model or queuing network, a modular modeling approach based on extended stochastic Petri nets (ESPN) is presented. The state space is got by building the embedded Markov chain based on the ESPN model. Relative excitation probability distribution of all processing statuses is inferred from station first pass yield. Therefore, the analysis of two-stage serial manufacturing system using Markov chain theory is presented. Furthermore, considering that both sides of system have the impact on the cell model, an overlapping decomposition method is employed for iterative solution of the model of multistage serial manufacturing system, and the performance index of the system are also discussed. An automotive powertrain assembly line is taken as an example to verify the effectiveness of the model and the validity of the analysis method.

Key words: extended-stochastic Petri nets, multistage serial manufacturing system, overlapping decomposition method, performance evaluation, product polymorphism, rework

中图分类号: