| 引用本文: | 邵炜世,皮德常,邵仲世.学习驱动的分布式异构混合流水车间批量流能效调度优化[J].控制理论与应用,2024,41(6):1018~1028.[点击复制] |
| SHAO Wei-shi,PI De-chang,SHAO Zhong-shi.Learning-driven optimization of energy-efficient distributed heterogeneous hybrid flow shop lot-streaming scheduling[J].Control Theory and Technology,2024,41(6):1018~1028.[点击复制] |
|
| 学习驱动的分布式异构混合流水车间批量流能效调度优化 |
| Learning-driven optimization of energy-efficient distributed heterogeneous hybrid flow shop lot-streaming scheduling |
| 摘要点击 613 全文点击 163 投稿时间：2022-07-16 修订日期：2022-12-06 |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| DOI编号 DOI: 10.7641/CTA.2023.20633 |
| 2024,41(6):1018-1028 |
| 中文关键词 分布式异构混合流水车间 批量流调度 学习驱动的多目标进化算法 整数规划 能效优化 |
| 英文关键词 distributed heterogeneous hybrid flow shop scheduling lot-streaming scheduling learning-driven multiobjective evolutionary algorithm integer programming energy-efficiency optimization |
| 基金项目 国家自然科学基金项目(62003203, 62103195, 62262018), 江苏省基础研究计划项目(BK20210558), 中国博士后基金面上项目(2021M701700, 2023M732166), 中央高校基本业务费项目(GK202201014), 大规模复杂系统数值模拟教育部重点实验室开放基金项目(202404)资助. |
|
| 中文摘要 |
| 本文研究了分布式异构混合流水车间批量流能效调度问题, 其中每个工厂的加工效率不同, 工件可以分割成若干子批进入加工系统. 以最大完成时间和总能耗为优化目标, 建立了混合整数规划模型. 本文提出了一种学习驱动的多目标进化算法, 包括学习驱动的全局搜索和局部搜索. 引入Q学习作为学习引擎, 以种群和非支配解集的评价作为环境反馈信号, 通过不断的学习来动态指导搜索操作的选择; 基于问题特征, 设计了算法的状态集、动作集和奖励机制. Q学习的引入能够及时感知当前搜索的状态, 减少搜索操作的盲目性, 提高搜索的效率. 通过对仿真数据集的测试, 表明所提出算法能够有效地求解分布式异构混合流水车间批量流能效调度问题. |
| 英文摘要 |
| This paper studies an energy-efficient distributed heterogeneous hybrid flow shop lot-streaming scheduling problem, where the processing efficiency of each factory is different and the jobs can be split into several sub-lots to access the manufacturing system. The mixed integer programming model is built with the makespan and total energy consumption objectives. A learning-driven multi-objective evolutionary algorithm is proposed, which includes learning-driven global search and local search. Q-learning is introduced as a learning engine, and the evaluation of population and non-dominated solution sets is used as an environmental feedback signal to dynamically guide the selection of search operations through continuous learning. Based on the characteristics of the problem, the state set, action set and reward mechanism of the algorithm are designed. The introduction of Q-learning can sense the current search state in time, reduce the blindness of search operations, and improve the efficiency of search. From the testing results on simulation data set, it is shown that the proposed algorithm can effectively solve the energy-efficient distributed heterogeneous hybrid flow shop lot-streaming scheduling problem. |
|
|
|
|
|