An ADMM-based parallel algorithm for solving traffic assignment problem with elastic demand

被引:7
作者
Zhang, Kai
Zhang, Honggang
Dong, Yu
Wu, Yunchi
Chen, Xinyuan [1 ,2 ]
机构
[1] Southeast Univ, Jiangsu Prov Collaborat Innovat Ctr Modern Urban T, Sch Transportat, Jiangsu Key Lab Urban ITS, Nanjing 211000, Peoples R China
[2] Nanjing Univ Aeronaut & Astronaut, Coll Civil Aviat, Nanjing 210016, Peoples R China
来源
COMMUNICATIONS IN TRANSPORTATION RESEARCH | 2023年 / 3卷
基金
中国国家自然科学基金;
关键词
Traffic assignment problem; Elastic demand; Model decomposition; Alternative direction method of multipliers; (ADMM); Parallel computing; USER EQUILIBRIUM; NETWORK EQUILIBRIUM; DECOMPOSITION; SPLIT;
D O I
10.1016/j.commtr.2023.100108
中图分类号
U [交通运输];
学科分类号
08 ; 0823 ;
摘要
Efficiently solving the user equilibrium traffic assignment problem with elastic demand (UE-TAPED) for transportation networks is a critical problem for transportation studies. Most existing UE-TAPED algorithms are designed using a sequential computing scheme, which cannot take advantage of advanced parallel computing power. Therefore, this study focuses on model decomposition and parallelization, proposing an origin-based formulation for UE-TAPED and proving an equivalent reformulation of the original problem. Furthermore, the alternative direction method of multipliers (ADMM) is employed to decompose the original problem into independent link-based subproblems, which can solve large-scale problems with small storage space. In addition, to enhance the efficiency of our algorithm, the parallel computing technology with optimal parallel computing schedule is implemented to solve the link-based subproblems. Numerical experiments are performed to validate the computation efficiency of the proposed parallel algorithm.
引用
收藏
页数:18
相关论文
共 71 条
  • [1] An efficient method to compute traffic assignment problems with elastic demands
    Babonneau, Frederic
    Vial, Jean-Philippe
    [J]. TRANSPORTATION SCIENCE, 2008, 42 (02) : 249 - 260
  • [2] Traffic assignment by paired alternative segments
    Bar-Gera, Hillel
    [J]. TRANSPORTATION RESEARCH PART B-METHODOLOGICAL, 2010, 44 (8-9) : 1022 - 1046
  • [3] Beckmann M., 1956, STUDIES EC TRANSPORT
  • [4] 2ND DERIVATIVE ALGORITHMS FOR MINIMUM DELAY DISTRIBUTED ROUTING IN NETWORKS
    BERTSEKAS, DP
    GAFNI, EM
    GALLAGER, RG
    [J]. IEEE TRANSACTIONS ON COMMUNICATIONS, 1984, 32 (08) : 911 - 919
  • [5] Distributed optimization and statistical learning via the alternating direction method of multipliers
    Boyd S.
    Parikh N.
    Chu E.
    Peleato B.
    Eckstein J.
    [J]. Foundations and Trends in Machine Learning, 2010, 3 (01): : 1 - 122
  • [6] Boyd S., 2004, Convex Optimization
  • [7] Brooks J, 2019, Arxiv, DOI arXiv:1509.08206
  • [9] Chen A., 1998, 77 ANN ME TRANSP RES
  • [10] Analytical formulation for explaining the variations in traffic states: A fundamental diagram modeling perspective with stochastic parameters
    Cheng, Qixiu
    Lin, Yuqian
    Zhou, Xuesong
    Liu, Zhiyuan
    [J]. EUROPEAN JOURNAL OF OPERATIONAL RESEARCH, 2024, 312 (01) : 182 - 197