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 ›› 2026, Vol. 34 ›› Issue (2): 79-88.doi: 10.16381/j.cnki.issn1003-207x.2024.0074cstr: 32146.14.j.cnki.issn1003-207x.2024.0074

• • 上一篇    下一篇

动态订单下无人机辅助骑手外卖配送路径优化研究

卢福强1(), 蒋润雪2, 毕华玲1, 高志远1   

  1. 1.燕山大学经济管理学院,河北 秦皇岛 066004
    2.南京航空航天大学经济与管理学院,江苏 南京 21106
  • 收稿日期:2024-01-12 修回日期:2024-06-24 出版日期:2026-02-25 发布日期:2026-02-04
  • 通讯作者: 卢福强 E-mail:fuqiang_lu@126.com
  • 基金资助:
    教育部人文社会科学研究项目(24YJC630129);河北省教育厅高等学校科技计划(理工类)项目(QN2025009);河北省自然科学基金项目G2024203003

Routing Optimization of Drone Assisted Riders Takeout Delivery under Dynamic Orders

Fuqiang Lu1(), Runxue Jiang2, Hualing Bi1, Zhiyuan Gao1   

  1. 1.School of Economics and Management,Yanshan University,Qinhuangdao 066004,China
    2.College of Economics and Management,Nanjing University of Aeronautics and Astronautics,Nanjing 211106,China
  • Received:2024-01-12 Revised:2024-06-24 Online:2026-02-25 Published:2026-02-04
  • Contact: Fuqiang Lu E-mail:fuqiang_lu@126.com

摘要:

针对外卖订单动态产生和骑手不断变化等问题,提出了无人机辅助骑手配送模式。利用双峰高斯函数模拟配送过程中新订单的生成,以最小配送成本和顾客整体最大满意度为目标函数构建了两阶段优化模型。其中,第一阶段针对静态顾客建立了骑手配送的初始优化模型,第二阶段针对动态产生的新订单,建立了无人机辅助骑手配送的动态调整模型。针对该模型设计了一个两阶段启发式算法进行求解,第一阶段采用基于K-means和KNN分类算法改进的AP聚类算法进行动态订单分配,第二阶段利用结合插入算法改进的禁忌搜索算法进行路径优化。通过算例仿真,验证模型和算法的有效性和可行性。结果表明,与传统的骑手配送模式相比,无人机辅助骑手配送模式能有效减少使用骑手数量、降低配送成本。

关键词: 无人机辅助骑手, 外卖配送, 动态路径优化, 禁忌搜索算法, AP聚类算法

Abstract:

Due to the time-sensitive nature of distribution orders, each order is required to be completed within the shortest possible time, and considering the characteristics of large number of distribution orders during peak periods, geographically dispersed, and dynamic and random distribution of riders, it is extremely challenging to seek an order allocation and route optimization scheme with low delivery cost and high customer satisfaction. In order to solve the problems of dynamic generation of takeout orders and the continuous change of riders, a drone assisted riders delivery mode is proposed. The bimodal Gaussian function is used to simulate the generation of new orders in the distribution process, and a two-stage optimization model is constructed with the minimum distribution cost and the overall maximum customer satisfaction as the objective functions. In the first stage, an initial optimization model for rider delivery is established for static customers, and in the second stage, a dynamic adjustment model for drone assisted rider delivery is established for dynamically generated new orders. A two-stage heuristic algorithm is designed for this model to solve the problem, the first stage uses an AP (Affinity propagation) clustering algorithm improved based on K-means and KNN (K-Nearest Neighbor) classification algorithms for dynamic order allocation, and the second stage utilizes a taboo search algorithm improved by combining the insertion algorithm for route optimization. The effectiveness and feasibility of the model and algorithm are verified through case simulation. The results show that:(1) Compared with the traditional rider delivery mode, the drone assisted riders delivery mode can effectively reduce the number of riders and reduce the delivery cost. And with the larger scale of new orders, the advantages of the drone assisted rider delivery mode become more obvious. (2) Through real-time adjustment of delivery routes and optimization of order allocation, the empty load rate and transportation costs can be reduced, and operational efficiency can be improved. (3) The addition of drones can effectively optimize the distribution path and schedule of orders, avoiding delays caused by traffic congestion or unreasonable routes. It helps to reduce operating costs and further improve the overall efficiency and service quality of the logistics industry.

Key words: drone assisted riders, takeout delivery, dynamic route optimization, taboo search algorithm, AP clustering algorithm

中图分类号: