主管:中国科学院
主办:中国优选法统筹法与经济数学研究会
   中国科学院科技战略咨询研究院

中国管理科学 ›› 2023, Vol. 31 ›› Issue (9): 148-158.doi: 10.16381/j.cnki.issn1003-207x.2021.1531

• • 上一篇    下一篇

数据驱动环境下考虑多重干扰情境的共享单车重置优化研究

刘明1, 徐锡芬1, 曹杰2()   

  1. 1.南京理工大学经济管理学院,江苏 南京 210094
    2.徐州工程学院管理工程学院,江苏 徐州 221018
  • 收稿日期:2021-08-04 修回日期:2021-09-22 出版日期:2023-09-15 发布日期:2025-02-12
  • 通讯作者: 曹杰 E-mail:cj@amss.ac.cn
  • 基金资助:
    国家自然科学基金资助项目(72171119)

A Data-driven Optimization Model for Reallocating Dock-less Sharing Bikes with Considering Multiple Disruption Scenarios

Ming LIU1, Xi-fen XU1, Jie CAO2()   

  1. 1.School of Economics and Management,Nanjing University of Science and Technology,Nanjing 210094,China
    2.School of Management Engineering,Xuzhou University of Technology,Xuzhou 221018,China
  • Received:2021-08-04 Revised:2021-09-22 Online:2023-09-15 Published:2025-02-12
  • Contact: Jie CAO E-mail:cj@amss.ac.cn

摘要:

无桩特性给共享单车重置调度带来了全新的问题和挑战,决策者面对的是随机分布在整个区域面上的单车如何调度,这与传统有桩公共自行车的重置有着显著区别。基于此,本文以某品牌共享单车为研究对象,通过分析其运行轨迹数据,界定出共享单车的活跃点(放车点)与非活跃点(收车点)。由于收车点的重置需求会因用户行为影响而不断发生变化,本文结合干扰管理的思想,综合考虑收车点的单车数变化、收车点消失以及新增收车点等干扰情境,建立考虑多重干扰情境的共享单车重置调度多目标优化模型,并设计带精英策略的非支配排序遗传算法对其求解。测试结果表明,本文所构建的模型和求解算法能够快速寻找到问题的帕累托最优解集,可为共享单车的重置调度提供有效的策略建议和决策支持。

关键词: 共享单车, 数据驱动, 干扰管理, 多目标优化, NSGA-II算法

Abstract:

The dock-less feature brings a new and challenging problem to reallocate the dock-less sharing bikes. Decision-makers need to consider how to reallocate these bikes, which are randomly parked over the entire area. This is significantly different from the traditional public bikes reallocating problem. Taking a brand bicycle as the research background, the active point (delivery point) and the inactive point (pick-up point) are defined by analyzing mass trajectory data of dock-less sharing bikes in this paper. Due to the status of pick-up points will be constantly changed, three scenarios are proposed to depict it, which include changing the number of bikes, disappearing the pick-up points, and pop-up new pick-up points. Based on the theory of disruption management, a multi-objective optimization model is proposed for reallocating the dock-less sharing bikes and a non-dominated sorting genetic algorithm II is present to solve it. The test result demonstrates that the proposed algorithm can quickly converge to the Pareto solution set of the problem, and thus it can provide effective suggestions and decision support for reallocating the dock-less sharing bikes.

Key words: dock-less sharing bike, data-driven, disruption management, multi-objective optimization, NSGA-II.

中图分类号: