Solving cyclic train timetabling problem through model reformulation

Extended time-space network construct and Alternating Direction Method of Multipliers methods

Yongxiang Zhang, Qiyuan Peng, Yu Yao, Xin Zhang, Xuesong Zhou

Research output: Contribution to journalArticle

Abstract

The cyclic train timetabling problem aims to synchronize limited operational resources toward a master periodic schedule of transport services. By introducing an extended time-space network construct, this paper proposes a new type of integer programming model reformulation for the cyclic train timetabling problem on a double-track railway corridor at the macroscopic level. This reformulation method also holds the promises to be applied in a broader set of routing and scheduling problems with periodic activity requirements. We also hope that this space-time network extension technique, as a special version of variable splitting methods in the dual decomposition literature, could potentially bridge the modeling gaps between cyclic and non-cyclic timetables. Specifically, the existing mathematical programming model for the periodic event scheduling problem (PESP) is transformed into a multi-commodity network flow model with two coupled schedule networks and side track capacity constraints. In addition, two dual decomposition methods including Lagrangian relaxation and Alternating Direction Method of Multipliers (ADMM), are adopted to dualize the side track capacity constraints. For each train-specific sub-problem in an iterative primal and dual optimization framework, we develop an enhanced version of forward dynamic programming to find the time-dependent least cost master schedule across the time-space network over multiple periods. ADMM-motivated heuristic methods with adjusted penalty parameters are also developed to obtain good upper bound solutions. Based on real-world instances from the Beijing-Shanghai high-speed railway corridor, we compare the numerical performance between the proposed reformulation and the PESP model that involves the standard optimization solver.

Original languageEnglish (US)
Pages (from-to)344-379
Number of pages36
JournalTransportation Research Part B: Methodological
Volume128
DOIs
StatePublished - Oct 1 2019

Fingerprint

multiplier
Scheduling
scheduling
programming
Decomposition
Heuristic methods
German Federal Railways
Mathematical programming
Integer programming
Dynamic programming
event
commodity
time
penalty
heuristics
Costs
costs
resources
performance

Keywords

  • ADMM
  • Cyclic train timetabling
  • Extended time-space network
  • Lagrangian relaxation

ASJC Scopus subject areas

  • Civil and Structural Engineering
  • Transportation

Cite this

@article{de95161917a14eb1b7b52e1db21b8eb6,
title = "Solving cyclic train timetabling problem through model reformulation: Extended time-space network construct and Alternating Direction Method of Multipliers methods",
abstract = "The cyclic train timetabling problem aims to synchronize limited operational resources toward a master periodic schedule of transport services. By introducing an extended time-space network construct, this paper proposes a new type of integer programming model reformulation for the cyclic train timetabling problem on a double-track railway corridor at the macroscopic level. This reformulation method also holds the promises to be applied in a broader set of routing and scheduling problems with periodic activity requirements. We also hope that this space-time network extension technique, as a special version of variable splitting methods in the dual decomposition literature, could potentially bridge the modeling gaps between cyclic and non-cyclic timetables. Specifically, the existing mathematical programming model for the periodic event scheduling problem (PESP) is transformed into a multi-commodity network flow model with two coupled schedule networks and side track capacity constraints. In addition, two dual decomposition methods including Lagrangian relaxation and Alternating Direction Method of Multipliers (ADMM), are adopted to dualize the side track capacity constraints. For each train-specific sub-problem in an iterative primal and dual optimization framework, we develop an enhanced version of forward dynamic programming to find the time-dependent least cost master schedule across the time-space network over multiple periods. ADMM-motivated heuristic methods with adjusted penalty parameters are also developed to obtain good upper bound solutions. Based on real-world instances from the Beijing-Shanghai high-speed railway corridor, we compare the numerical performance between the proposed reformulation and the PESP model that involves the standard optimization solver.",
keywords = "ADMM, Cyclic train timetabling, Extended time-space network, Lagrangian relaxation",
author = "Yongxiang Zhang and Qiyuan Peng and Yu Yao and Xin Zhang and Xuesong Zhou",
year = "2019",
month = "10",
day = "1",
doi = "10.1016/j.trb.2019.08.001",
language = "English (US)",
volume = "128",
pages = "344--379",
journal = "Transportation Research, Series B: Methodological",
issn = "0191-2615",
publisher = "Elsevier Limited",

}

TY - JOUR

T1 - Solving cyclic train timetabling problem through model reformulation

T2 - Extended time-space network construct and Alternating Direction Method of Multipliers methods

AU - Zhang, Yongxiang

AU - Peng, Qiyuan

AU - Yao, Yu

AU - Zhang, Xin

AU - Zhou, Xuesong

PY - 2019/10/1

Y1 - 2019/10/1

N2 - The cyclic train timetabling problem aims to synchronize limited operational resources toward a master periodic schedule of transport services. By introducing an extended time-space network construct, this paper proposes a new type of integer programming model reformulation for the cyclic train timetabling problem on a double-track railway corridor at the macroscopic level. This reformulation method also holds the promises to be applied in a broader set of routing and scheduling problems with periodic activity requirements. We also hope that this space-time network extension technique, as a special version of variable splitting methods in the dual decomposition literature, could potentially bridge the modeling gaps between cyclic and non-cyclic timetables. Specifically, the existing mathematical programming model for the periodic event scheduling problem (PESP) is transformed into a multi-commodity network flow model with two coupled schedule networks and side track capacity constraints. In addition, two dual decomposition methods including Lagrangian relaxation and Alternating Direction Method of Multipliers (ADMM), are adopted to dualize the side track capacity constraints. For each train-specific sub-problem in an iterative primal and dual optimization framework, we develop an enhanced version of forward dynamic programming to find the time-dependent least cost master schedule across the time-space network over multiple periods. ADMM-motivated heuristic methods with adjusted penalty parameters are also developed to obtain good upper bound solutions. Based on real-world instances from the Beijing-Shanghai high-speed railway corridor, we compare the numerical performance between the proposed reformulation and the PESP model that involves the standard optimization solver.

AB - The cyclic train timetabling problem aims to synchronize limited operational resources toward a master periodic schedule of transport services. By introducing an extended time-space network construct, this paper proposes a new type of integer programming model reformulation for the cyclic train timetabling problem on a double-track railway corridor at the macroscopic level. This reformulation method also holds the promises to be applied in a broader set of routing and scheduling problems with periodic activity requirements. We also hope that this space-time network extension technique, as a special version of variable splitting methods in the dual decomposition literature, could potentially bridge the modeling gaps between cyclic and non-cyclic timetables. Specifically, the existing mathematical programming model for the periodic event scheduling problem (PESP) is transformed into a multi-commodity network flow model with two coupled schedule networks and side track capacity constraints. In addition, two dual decomposition methods including Lagrangian relaxation and Alternating Direction Method of Multipliers (ADMM), are adopted to dualize the side track capacity constraints. For each train-specific sub-problem in an iterative primal and dual optimization framework, we develop an enhanced version of forward dynamic programming to find the time-dependent least cost master schedule across the time-space network over multiple periods. ADMM-motivated heuristic methods with adjusted penalty parameters are also developed to obtain good upper bound solutions. Based on real-world instances from the Beijing-Shanghai high-speed railway corridor, we compare the numerical performance between the proposed reformulation and the PESP model that involves the standard optimization solver.

KW - ADMM

KW - Cyclic train timetabling

KW - Extended time-space network

KW - Lagrangian relaxation

UR - http://www.scopus.com/inward/record.url?scp=85071140540&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85071140540&partnerID=8YFLogxK

U2 - 10.1016/j.trb.2019.08.001

DO - 10.1016/j.trb.2019.08.001

M3 - Article

VL - 128

SP - 344

EP - 379

JO - Transportation Research, Series B: Methodological

JF - Transportation Research, Series B: Methodological

SN - 0191-2615

ER -