计算机集成制造系统 ›› 2021, Vol. 27 ›› Issue (5): 1351-1360.DOI: 10.13196/j.cims.2021.05.011

• 当期目次 • 上一篇    下一篇

基于SO-GP的智能车间组合调度规则挖掘

马丽萌,乔非,马玉敏+,刘鹃   

  1. 同济大学电子与信息工程学院
  • 出版日期:2021-05-31 发布日期:2021-05-31
  • 基金资助:
    国家自然科学基金资助项目(71690230/71690234,61873191,61973237);国家重点研发计划资助项目(2017YFE0101400)。

Mining of composite dispatching rules for smart workshop based on SO-GP

  • Online:2021-05-31 Published:2021-05-31
  • Supported by:
    Project supported by the National Natural Science Foundation,China(No.71690230/71690234,61873191,61973237),and the National Key Research & Development Program,China(No.2017YFE0101400).

摘要: 针对智能车间复杂性程度高、动态不确定性明显、对调度的实时性要求高以及车间机理模型难以描述等特点,对智能车间实时调度问题展开研究,提出一种将仿真优化与遗传规划(SO-GP)算法相结合的调度规则挖掘方法,在优化生产性能的同时满足实时调度的需求。在SO-GP方法设计中,采用二叉树的结构编码,每个GP个体代表一个组合调度规则,并借助仿真获得个体的适应度值。为了进一步提高挖掘过程的时间效率,对构成GP算法的终止集进行了归一化改进。最后以MiniFAB半导体生产线模型为对象进行实验,验证了所提方法的有效性。

关键词: 智能车间, 生产调度, 调度规则, GP算法, 仿真优化

Abstract: To study the real-time scheduling of the smart workshop,Simulation Optimization-Genetic Programming (SO-GP) algorithm was proposed to mine the dispatching rules for smart workshops with high complexity,obvious dynamic uncertainties,high real-time dispatching requirements and difficulties of describing the workshop's mechanism model,which could meet the requirements of real-time scheduling while optimizing production performance.In SO-GP method,the structure coding of binary tree was used.Each GP individual represented a combined scheduling rule,and the fitness value of the individual was obtained through simulation.To further improve the time efficiency of mining process,the termination set that had constituted the GP algorithm was normalized and improved.The MiniFAB semiconductor production line was used as the object,and the experimental verifications were carried out under different dispatching aims.Experiments showed that the proposed method was able to improve the production performance of the smart shop by comparing with the traditional dispatching rules.

Key words: smart workshop, production dispatching, dispatching rules, genetic programming algorithm, simulation optimization

中图分类号: