基于向量映射代理模型的分布式柔性作业车间调度算法
CSTR:
作者:
作者单位:

北京航空航天大学

作者简介:

通讯作者:

中图分类号:

TH165; TP181

基金项目:

卓越工程师产教联合培养研究基地卓越工程师培养改革战略研究课题


Distributed Flexible Job Shop Scheduling Algorithm Based on a Vector Mapping Surrogate Model
Author:
Affiliation:

Beihang University

Fund Project:

Excellent engineer training reform strategy research topic of production and education joint training research base

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    随着定制化制造需求的增加,分布式柔性作业调度问题涉及的调度任务复杂多变,多产线制造设备动态重构,人机料法环协同约束倍增.面对多场景任务调度需求,单一进化算法难以适应不同场景,且由于算法对这类复杂约束问题评估时间成本较高,逐场景调用算法生成调度方案十分耗时.为解决这一问题,本文建立了分布式柔性作业车间调度问题数学模型,并提出基于向量映射代理模型(VMSM)的调度算法.VMSM通过高维特征向量映射进行特征提取,以增强对历史相似问题的识别能力,并在进化算法中辅助生成初始解和评估解集,提高搜索效率.实验结果表明,该方法能够将相似调度方案的分类准确率提升25\%-35\%,有效减少进化算法初始解筛选和解集评估次数.在不同场景复杂度下,VMSM均能在保障调度方案质量的前提下,显著提高分布式柔性作业车间调度方案生成速度,方案生成时间最多缩短51.26\%.

    Abstract:

    As the demand for customized manufacturing grows, the Distributed Flexible Job Shop Scheduling Problem (DFJSP) involves complex,variable scheduling tasks, dynamic reconfiguration of multi-line manufacturing equipment, and increased collaborative constraints across human, machine, material, method, and environment. Addressing the multi-scenario scheduling requirements is challenging, as single evolutionary algorithms struggle to adapt to diverse scenarios and incur high time costs when assessing such complex constraints. To address this issue, this paper establishes a mathematical model for the DFJSP and subsequently proposes a scheduling algorithm based on the Vector Mapping Surrogate Model (VMSM). VMSM enhances the recognition of historically similar problems through high-dimensional feature vector mapping and aids in generating initial solutions and evaluating solution sets within evolutionary algorithms, thereby improving search efficiency. Experimental results demonstrate that this method increases classification accuracy for similar scheduling solutions by 25\%-35\%, significantly reducing the number of initial solution selections and solution set evaluations. Under various scenario complexities, VMSM effectively enhances DFJSP solution generation speed by up to 51.26\%, while maintaining the quality of the scheduling solutions.

    参考文献
    相似文献
    引证文献
引用本文
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2024-08-14
  • 最后修改日期:2024-11-13
  • 录用日期:2024-11-14
  • 在线发布日期: 2024-11-22
  • 出版日期:
文章二维码