计算机工程与应用 ›› 2012, Vol. 48 ›› Issue (9): 12-14.

• 博士论坛 • 上一篇    下一篇

干扰环境下多产品循环制造网络自适应技术

蔡政英1,2,王燕舞1,肖人彬1,肖江文1   

  1. 1.华中科技大学 系统工程研究所,武汉 430074
    2.三峡大学 计算机与信息学院,湖北 宜昌 443002
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2012-03-21 发布日期:2012-04-11

Adaptive technology of multi-product closed-loop manufacturing network under disruption environment

CAI Zhengying1,2, WANG Yanwu1, XIAO Renbin1, XIAO Jiangwen1   

  1. 1.Institute of Systems Engineering, Huazhong University of Science and Technology, Wuhan 430074, China
    2.College of Computer and Information Technology, China Three Gorges University, Yichang, Hubei 443002, China
  • Received:1900-01-01 Revised:1900-01-01 Online:2012-03-21 Published:2012-04-11

摘要: 针对多产品循环制造链协同运作问题,采用多智能体进行建模。根据供应链运作参考模型,将循环制造链建模为包括供应商智能体、制造商智能体、分销商智能体等多种智能体的网络模型,根据不同功能特点在内部进一步分解为采购、制造、分销、再采购、再制造、再分销等子智能体,分析了这些智能体运作中的干扰问题。根据多智能体系统干扰的不确定性和交互性,建立一种神经网络自学习机制,能够不依赖于系统的初始设置参数而实现多智能体在线自学习和自调整,给出了算法的求解步骤。用仿真和实验验证了该方案的可行性。

关键词: 多智能体, 循环制造链, 干扰优化, 神经网络

Abstract: According to coordination problem of multi-product closed-loop manufacturing network, a multi-agent model is proposed. Based on supply chain operations reference model, the closed-loop supply chain is modeled as a multi-agent network topology made up of supplier agent, manufacturer agent and distributor agent, and their subagents are further divided into source, make, distribute, re-resource, re-make, re-distribute, whose disruption problem is also analyzed. According to uncertainties and interactions between agent disruptions, a neural network self-learning mechanism is presented which can self-learn and self-adjust the control parameters of multi-agent system online instead of depending on initialized system parameters, and its solving steps are given as well. The proposed method is verified by simulation and experimental results.

Key words: multi-agent, closed-loop manufacturing chain, disruption optimization, neural network