中国机械工程

• 智能制造 • 上一篇    下一篇

基于混合教学优化算法的多车间协作综合调度

廖不凡1;雷琦1;吴文烈2;宋豫川1;郭伟飞1   

  1. 1.重庆大学机械传动国家重点实验室,重庆,400044
    2.复旦大学科学技术研究院,上海,200433
  • 出版日期:2020-08-25 发布日期:2020-09-17
  • 基金资助:
    国家自然科学基金资助项目(512045429);
    国家工信部资助项目(CDGC01-KT0603)

Hybrid Teaching-learning-based Optimization Algorithms for Integrated Scheduling of Multi-workshop Collaborations

LIAO Bufan1;LEI Qi1;WU Wenlie2;SONG Yuchuan1;GUO Weifei1   

  1. 1.State Key Laboratory of Mechanical Transmission,Chongqing University,Chongqing,400044
    2.Institute of Science and Technology,Fudan University,Shanghai,200433
  • Online:2020-08-25 Published:2020-09-17

摘要: 目前在处理加工属性类似的多车间协作综合调度问题时,几乎都是采用调度规则,对产品加工工艺结构的依赖度过高,降低了算法对同一类问题不同实例的适应性,并且对大型实例的求解结果普遍欠佳,为此提出了一种混合教学优化算法。该算法在基本教学优化算法的基础上加入采用变异操作模拟的自学习阶段,提高其局部搜索能力,并且在教学、互学和自学三个阶段均按照模拟退火算法中Metropolis准则计算的概率,随机接受学生群体中某一个较差个体作为新个体,进一步提高算法跳出局部最优解的能力;教学、互学和自学三个阶段设计的变换操作均考虑综合调度问题中各虚拟工序之间的顺序约束关系,保证生成的解均是可行解。通过测试以往该类问题实例,得到的结果验证了所提算法的可行性和有效性。

关键词: 多车间协作, 综合调度, 教学优化算法, 自学习, Metropolis准则

Abstract: Aiming at the current integrated scheduling problems of multi-workshop collaborations with similar processing properties, where almost using the scheduling rules, relying too heavily on the processing structures of the products, reducing the algorithm adaptability to different instances on the same type of problems, and the poor results of solving large instances, a hybrid teaching-learning-based optimization algorithm was proposed. A self-learning phase was added, the mutation operators were executed in the individual self-learning phase to improve the local search ability. In the three phases of teaching, mutual-learning and self-learning, the probability calculated by the Metropolis procedure in the simulated annealing algorithm was adopted to accept randomly a bad individual as a new one, further improving the ability of running away from local optima. The transformation operations designed in the three phases taken into account the sequential constraint relations among the virtual processes in the integrated scheduling problems and ensure that the solutions generated were all feasible solutions. The feasibility and validity of the proposed algorithm were verified by testing the previous examples.

Key words: multi-workshop collaboration, integrated scheduling, teaching-learning-based optimization algorithm, self-learning, Metropolis procedure

中图分类号: