运筹与管理 ›› 2025, Vol. 34 ›› Issue (12): 78-84.DOI: 10.12005/orms.2025.0378

• 理论分析与方法探讨 • 上一篇    下一篇

带有学习效应的多窗口分配调度问题研究

刘峥1, 吕丹阳1, 冯伟2, 王吉波1   

  1. 1.沈阳航空航天大学 经济与管理学院,辽宁 沈阳 110136;
    2.沈阳航空航天大学 图书馆,辽宁 沈阳 110136
  • 收稿日期:2024-08-02 出版日期:2025-12-25 发布日期:2026-04-29
  • 通讯作者: 王吉波(1975-),男,辽宁沈阳人,博士,教授,博士生导师,研究方向:运筹学与组合优化。Email: wangjibo75@163.com。
  • 作者简介:刘峥(2000-),男,辽宁锦州人,硕士研究生,研究方向:物流管理与运筹学。
  • 基金资助:
    国家自然科学基金资助项目(71471120)
       

Research on Multiple Due-window Assignment Scheduling Problems with Learning Effects

LIU Zheng1, LYU Danyang1, FENG Wei2, WANG Jibo1   

  1. 1. School of Economics and Management, Shenyang Aerospace University, Shenyang 110136, China;
    2. Department of Library, Shenyang Aerospace University, Shenyang 110136, China
  • Received:2024-08-02 Online:2025-12-25 Published:2026-04-29

摘要: 本文在两种经典的工期窗口(共同工期窗口和松弛工期窗口)下,考虑给定工件具有多个共同工期窗口或松弛流的问题,将工件依据其特性划分为不同组,其中同一组内的工件共享相同的工期窗口或松弛流参数,且工件的加工时间并非固定不变,而是与学习效应紧密相关。核心目标为寻找一个最优的工件加工顺序以及确定多个共同工期窗口或松弛流的配置,以极小化工件的提前-延误成本、窗口开始时间/共同松弛流,以及窗口大小的线性加权和。通过理论分析与模型构建,当工件的分组数量确定时,该调度问题可以转化为一个指派问题,其时间复杂度为,且为给定的工件数。

关键词: 调度, 学习效应, 共同工期窗口, 共同松弛流, 提前-延误成本, 指派问题

Abstract: This paper provides insights into a specific and complex class of single-machine scheduling problems in the framework of two classical due-window settings i.e., common due-window and slack due-window. These problems involve jobs that are given multiple common due-windows or slack due-windows, where the jobs are divided into different groups based on their characteristics, and the jobs within the same group share the common due-window or slack due-window. In particular, the processing times of these jobs are not fixed, but are closely related to the learning effect, i.e., along with processing, the improvement of the worker’s skill or the optimization of the machine’s state leads to a reduction in the processing time of the subsequent jobs.
The core objective of the study is to find an optimal job processing sequence and determine multiple configurations of the common due-window or slack due-window, where the goal is to minimize the combined effect of several key performance metrics: earliness-tardiness costs for jobs, the cost of adjustments to the due-window start time, the efficiency of utilizing the common/slack due-window and the optimization of the size of the due-window. These metrics are considered together in the form of a linear weighted sum, aiming to balance the complex relationship among productivity, resource allocation and cost control.
Through theoretical analysis and model construction, it is found that the scheduling problem can be skillfully transformed into an assignment problem when the number of groupings of jobs is given, and then the problem can be solved in polynomial time. In this transformation process, we construct the corresponding mathematical model by exploiting the properties of the learning effect, the flexibility of the due-window and the regulating ability of the slack due-window.
Further, the Hungarian algorithm is used to solve the assignment problem where the time complexity is O(n3), where n is the number of jobs. This result not only provides a feasible path for solving this kind of complex single-machine scheduling problems, but also reveals the trend of the algorithm efficiency when the problem size (i.e., the number of jobs) increases, which provides a solid theoretical foundation for decision support in practical applications.
In summary, this study not only deepens the understanding of single-machine scheduling problems with learning effects and complex common/slack due-window constraints, but also provides effective solutions for efficiency improvement and cost optimization in real production scheduling through theoretical analysis and algorithms design.

Key words: scheduling, learning effect, common due-window, common flow-allowance, earliness-tardiness cost, assignment problem

中图分类号: