A Hybrid of Deep Reinforcement Learning and Local Search for the Vehicle Routing Problems

被引:87
|
作者
Zhao, Jiuxia [1 ]
Mao, Minjia [2 ]
Zhao, Xi [3 ]
Zou, Jianhua [1 ]
机构
[1] Xi An Jiao Tong Univ, Sch Elect & Informat Engn, Xian 710049, Peoples R China
[2] Xi An Jiao Tong Univ, Sch Math & Stat, Xian 710049, Peoples R China
[3] Xi An Jiao Tong Univ, Sch Management, Xian 710049, Peoples R China
基金
中国国家自然科学基金;
关键词
Routing; Adaptation models; Heuristic algorithms; Search problems; Training; Optimization; VRP; VRPTW; routing simulator; deep reinforcement learning; adaptive critic; local search; LARGE NEIGHBORHOOD SEARCH; OPTIMIZATION; ALGORITHMS; DELIVERY;
D O I
10.1109/TITS.2020.3003163
中图分类号
TU [建筑科学];
学科分类号
0813 ;
摘要
Different variants of the Vehicle Routing Problem (VRP) have been studied for decades. State-of-the-art methods based on local search have been developed for VRPs, while still facing problems of slow running time and poor solution quality in the case of large problem size. To overcome these problems, we first propose a novel deep reinforcement learning (DRL) model, which is composed of an actor, an adaptive critic and a routing simulator. The actor, based on the attention mechanism, is designed to generate routing strategies. The adaptive critic is devised to change the network structure adaptively, in order to accelerate the convergence rate and improve the solution quality during training. The routing simulator is developed to provide graph information and reward with the actor and adaptive cirtic. Then, we combine this DRL model with a local search method to further improve the solution quality. The output of the DRL model can serve as the initial solution for the following local search method, from where the final solution of the VRP is obtained. Tested on three datasets with customer points of 20, 50 and 100 respectively, experimental results demonstrate that the DRL model alone finds better solutions compared to construction algorithms and previous DRL approaches, while enabling a 5- to 40-fold speedup. We also observe that combining the DRL model with various local search methods yields excellent solutions at a superior generation speed, comparing to that of other initial solutions.
引用
收藏
页码:7208 / 7218
页数:11
相关论文
共 50 条
  • [31] Deep reinforcement learning for the dynamic and uncertain vehicle routing problem
    Pan, Weixu
    Liu, Shi Qiang
    APPLIED INTELLIGENCE, 2023, 53 (01) : 405 - 422
  • [32] Deep Policy Dynamic Programming for Vehicle Routing Problems
    Kool, Wouter
    van Hoof, Herke
    Gromicho, Joaquim
    Welling, Max
    INTEGRATION OF CONSTRAINT PROGRAMMING, ARTIFICIAL INTELLIGENCE, AND OPERATIONS RESEARCH, CPAIOR 2022, 2022, 13292 : 190 - 213
  • [33] Learning Improvement Heuristics for Solving Routing Problems..
    Wu, Yaoxin
    Song, Wen
    Cao, Zhiguang
    Zhang, Jie
    Lim, Andrew
    IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, 2022, 33 (09) : 5057 - 5069
  • [34] A hybrid metaheuristic algorithm based on iterated local search for vehicle routing problem with simultaneous pickup and delivery
    Oztas, Tayfun
    Tus, Aysegul
    EXPERT SYSTEMS WITH APPLICATIONS, 2022, 202
  • [35] A hybrid search method for the vehicle routing problem with time windows
    Brandao de Oliveira, Humberto Cesar
    Vasconcelos, Germano Crispim
    ANNALS OF OPERATIONS RESEARCH, 2010, 180 (01) : 125 - 144
  • [36] Deep Reinforcement Learning for Combinatorial Optimization: Covering Salesman Problems
    Li, Kaiwen
    Zhang, Tao
    Wang, Rui
    Wang, Yuheng
    Han, Yi
    Wang, Ling
    IEEE TRANSACTIONS ON CYBERNETICS, 2022, 52 (12) : 13142 - 13155
  • [37] Joint Optimization of Microservice Deployment and Routing in Edge via Multi-Objective Deep Reinforcement Learning
    Hu, Menglan
    Wang, Hao
    Xu, Xiaohui
    He, Jianwen
    Hu, Yi
    Deng, Tianping
    Peng, Kai
    IEEE TRANSACTIONS ON NETWORK AND SERVICE MANAGEMENT, 2024, 21 (06): : 6364 - 6381
  • [38] Meta-Learning-Based Deep Reinforcement Learning for Multiobjective Optimization Problems
    Zhang, Zizhen
    Wu, Zhiyuan
    Zhang, Hang
    Wang, Jiahai
    IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, 2023, 34 (10) : 7978 - 7991
  • [39] A Review of Heuristics and Hybrid Methods for Green Vehicle Routing Problems considering Emissions
    Gil, Alejandro Fernandez
    Lalla-Ruiz, Eduardo
    Sanchez, Mariam Gomez
    Castro, Carlos
    JOURNAL OF ADVANCED TRANSPORTATION, 2022, 2022
  • [40] VARL: a variational autoencoder-based reinforcement learning Framework for vehicle routing problems
    Wang, Qi
    APPLIED INTELLIGENCE, 2022, 52 (08) : 8910 - 8923