中国机械工程 ›› 2020, Vol. 31 ›› Issue (14): 1733-1740.DOI: 10.3969/j.issn.1004-132X.2020.14.013

• 智能制造 • 上一篇    下一篇

带时间窗的多中心半开放式车辆路径问题

辜勇, 袁源乙, 张列, 段晶晶   

  1. 武汉理工大学物流工程学院, 武汉, 430063
  • 收稿日期:2019-09-20 出版日期:2020-07-25 发布日期:2020-08-26
  • 通讯作者: 袁源乙(通信作者),女,1994年生,硕士研究生。研究方向为物流系统分析与优化。E-mail:460915419@qq.com。
  • 作者简介:辜勇,男,1975年生,副教授、博士。研究方向为系统分析与优化、智能物流规划与仿真。E-mail:guyong@whut.edu.cn。
  • 基金资助:
    国家重点研发计划资助项目(2018YFC1407405)

Multi-depot Half Open Vehicle Routing Problem with Time Windows

GU Yong, YUAN Yuanyi, ZHANG Lie, DUAN Jingjing   

  1. School of Logistics Engineering, Wuhan University of Technology, Wuhan, 430063
  • Received:2019-09-20 Online:2020-07-25 Published:2020-08-26

摘要: 针对多中心协同配送下的车辆路径问题,建立了总成本最小化模型,所建模型满足多中心、多需求点和半开放式的特征。考虑到问题的复杂性,设计了一种三阶段求解算法:将K-mediods聚类算法用于原始数据分解,将原规模较大的多配送中心路径问题转换成多个单配送中心路径问题;设计了改进多蚁群算法来求解单配送中心路径问题,得到初始方案;在调整阶段,利用节约算法优化初始方案。分析了算例,并同其他文献的算法求解结果进行对比,结果表明,所提算法比GA-ACO算法求解得到的单中心配送最优路径值减小32.16%,总成本减小30.42%;比狼群算法解得的最优路径值和总成本均减小8.99%;比蚁群算法求得的最优路径值减小24.76%,最小配送成本减小3.40%,从而验证了所建模型的合理性和所设计多阶段算法的有效性。

关键词: 多中心车辆路径问题, 协同配送, 时间窗, K-mediods聚类, 多蚁群算法

Abstract: Aiming at the vehicle routing problem under multi-depot collaborative distribution, a minimization model of total costs was established. The model satisfied the characteristics of multi-depot, multi-demand points and half-open. Considering the complexity of the problem, a three-stage solution algorithm was designed. K-mediods clustering algorithm was used to decompose the original data. The original large-scale multi-depot VRP was transformed into multiple single distribution center routing problems. Then, an improved multi-ant colony algorithm was designed to solve the single distribution center routing problem, and the initial scheme was obtained. In the adjustment stages, the initial scheme was optimized by using the saving mileage method. At last, the simulation results were compared with ones of the other methods. The results show that the proposed algorithm is better than GA-ACO algorithm, wolf pack algorithm and ant colony algorithm. The optimal path is optimized by 32.16%, 8.99% and 24.76% respectively, and the total cost is optimized by 30.42%, 8.99% and 3.40% separately, which verifies the rationality of the model and the effectiveness of the multi-stage algorithm design.

Key words: multi-depot vehicle routing problem(VRP), collaborative distribution, time windows, K-mediods clustering, multi-ant colony algorithm

中图分类号: