工业工程 ›› 2021, Vol. 24 ›› Issue (2): 110-118.doi: 10.3969/j.issn.1007-7375.2021.02.014

• 实践与应用 • 上一篇    下一篇

求解一类无关并行机调度的遗传迭代贪心算法

曾创锋, 刘建军, 陈庆新, 毛宁   

  1. 广东工业大学 广东省计算机集成制造重点实验室,广东 广州 510006
  • 收稿日期:2019-10-28 发布日期:2021-04-25
  • 通讯作者: 刘建军,刘建军(1982-),男,江西省人,副教授,主要研究方向为工业工程、生产计划与控制等。E-mail:jianjun.liu@gdut.edu.cn E-mail:jianjun.liu@gdut.edu.cn
  • 作者简介:曾创锋(1995-),男,广东省人,硕士研究生,主要研究方向为智能优化算法、生产调度等
  • 基金资助:
    国家自然科学基金资助项目(51975129, 71572049, 61973089);广东省自然科学基金资助项目(2019A1515012158);广东省特支计划科技创新青年拔尖人才项目(2016TQ03X364)

A Genetic Algorithm-Iterative Greedy Algorithm for a Kind of Unrelated Parallel Machine Scheduling Problem

ZENG Chuangfeng, LIU Jianjun, CHEN Qingxin, MAO Ning   

  1. Guangdong CIM Provincial Key Lab, Guangdong University of Technology, Guangzhou 510006, China
  • Received:2019-10-28 Published:2021-04-25

摘要: 以最小化最大完工时间为优化目标,建立带工单加工约束和序相关设置时间无关并行机调度问题的混合整数规划模型;考虑现实生产对求解算法在质量、收敛速度和鲁棒性等方面的较高要求,构建一种混合遗传-迭代贪心算法。在遗传变异操作中嵌入一种迭代贪心策略的破坏和构建机制,用于提高算法的种群多样性;引入基于破坏与构建操作设计而成的快速局部搜索算法来增强算法的局部开发能力;基于实际生产数据的相关特征随机生成了一系列计算案例,并通过实验说明所提新型混合算法相较于传统混合算法的优越性。

关键词: 无关并行机调度, 序相关设置时间, 遗传算法, 迭代贪心策略

Abstract: A deliberation is conducted on the unrelated parallel machine scheduling problem with job processing constraints and sequence-dependent setup times (UPMSP_JPCSST), that is, the same job has a number of alternative heterogeneous machines for processing, and the setup time of adjacent jobs on the same machine is determined by the product type and material type attributes. The objective is to minimize the total completion time and formulates a mixed integer programming model. Considering the higher requirements of practical production on the quality, convergence speed and robustness of the solution algorithm, a hybrid algorithm (genetic algorithm-iterated greedy algorithm, GAIG) is constructed. Firstly, the destruction and construction mechanism of an iterative greedy strategy is embedded in the genetic mutation operation to improve the population diversity of the algorithm. Secondly, a fast local search algorithm based on destruction and construction operation is introduced to enhance the local development ability of the algorithm; and finally, a series of calculation cases are generated randomly based on the relevant characteristics of actual production data, and the advantages of the proposed new hybrid algorithm over the traditional hybrid algorithm are illustrated through experiments.

Key words: unrelated parallel machine scheduling, sequence-dependent setup times, genetic algorithm, iterative greedy strategy

中图分类号: