Adaptive traffic control for large-scale dynamic traffic assignment applications

Alexander Paz, Yi-Chang Chiu

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

Dynamic traffic assignment (DTA) applications require traffic signal control data that are typically difficult to obtain and cumbersome to code in the required format. In addition, the evaluation of future scenarios requires future traffic signal settings consistent with the forecast demand. These signal settings are not known a priori and are costly to estimate. Intuitively, future signal timings need to be reasonably optimized so as to represent what the traffic management agency will do. In the literature, integration between traffic control and DTA models has been formulated as a bi-level or single-level optimization problem with system or user optimal constraints. Most existing solution procedures require certain nested structure with an inner-loop algorithm solving the problem of user-equilibrium or system-optimal assignment and the outer-loop algorithm searching for the optimal signal-timing settings. Most of these solution approaches remain only research tools without practical use because of computational intractability. This research proposes an efficient solution algorithm to the problem. An adaptive traffic signal control model is embedded in a simulation-based DTA model. For each inbound approach at an intersection of interest, the adaptive model uses upstream information and a dynamic rolling-horizon approach to project traffic flow conditions for a dynamic but short (projection) period. The adaptive model provides the signal settings during the entire process of traffic flow simulation and for every iteration of the solution algorithm. Thus, during the entire solution process, the experienced travel times and resulting traffic assignment flows are based on the adaptive (demand-responsive) signal settings, allowing the DTA flows and the adaptive signal settings to be generated simultaneously in a single-loop algorithmic structure. Simulation experiments illustrate the capabilities of the proposed approach.

Original languageEnglish (US)
Pages (from-to)103-112
Number of pages10
JournalTransportation Research Record
Issue number2263
DOIs
StatePublished - Dec 1 2011

Fingerprint

Traffic control
Traffic signals
Optimal systems
Information use
Flow simulation
Travel time
Experiments

ASJC Scopus subject areas

  • Civil and Structural Engineering
  • Mechanical Engineering

Cite this

Adaptive traffic control for large-scale dynamic traffic assignment applications. / Paz, Alexander; Chiu, Yi-Chang.

In: Transportation Research Record, No. 2263, 01.12.2011, p. 103-112.

Research output: Contribution to journalArticle

@article{d89270e7d8dc423b88b5ea09b1295029,
title = "Adaptive traffic control for large-scale dynamic traffic assignment applications",
abstract = "Dynamic traffic assignment (DTA) applications require traffic signal control data that are typically difficult to obtain and cumbersome to code in the required format. In addition, the evaluation of future scenarios requires future traffic signal settings consistent with the forecast demand. These signal settings are not known a priori and are costly to estimate. Intuitively, future signal timings need to be reasonably optimized so as to represent what the traffic management agency will do. In the literature, integration between traffic control and DTA models has been formulated as a bi-level or single-level optimization problem with system or user optimal constraints. Most existing solution procedures require certain nested structure with an inner-loop algorithm solving the problem of user-equilibrium or system-optimal assignment and the outer-loop algorithm searching for the optimal signal-timing settings. Most of these solution approaches remain only research tools without practical use because of computational intractability. This research proposes an efficient solution algorithm to the problem. An adaptive traffic signal control model is embedded in a simulation-based DTA model. For each inbound approach at an intersection of interest, the adaptive model uses upstream information and a dynamic rolling-horizon approach to project traffic flow conditions for a dynamic but short (projection) period. The adaptive model provides the signal settings during the entire process of traffic flow simulation and for every iteration of the solution algorithm. Thus, during the entire solution process, the experienced travel times and resulting traffic assignment flows are based on the adaptive (demand-responsive) signal settings, allowing the DTA flows and the adaptive signal settings to be generated simultaneously in a single-loop algorithmic structure. Simulation experiments illustrate the capabilities of the proposed approach.",
author = "Alexander Paz and Yi-Chang Chiu",
year = "2011",
month = "12",
day = "1",
doi = "10.3141/2263-12",
language = "English (US)",
pages = "103--112",
journal = "Transportation Research Record",
issn = "0361-1981",
publisher = "US National Research Council",
number = "2263",

}

TY - JOUR

T1 - Adaptive traffic control for large-scale dynamic traffic assignment applications

AU - Paz, Alexander

AU - Chiu, Yi-Chang

PY - 2011/12/1

Y1 - 2011/12/1

N2 - Dynamic traffic assignment (DTA) applications require traffic signal control data that are typically difficult to obtain and cumbersome to code in the required format. In addition, the evaluation of future scenarios requires future traffic signal settings consistent with the forecast demand. These signal settings are not known a priori and are costly to estimate. Intuitively, future signal timings need to be reasonably optimized so as to represent what the traffic management agency will do. In the literature, integration between traffic control and DTA models has been formulated as a bi-level or single-level optimization problem with system or user optimal constraints. Most existing solution procedures require certain nested structure with an inner-loop algorithm solving the problem of user-equilibrium or system-optimal assignment and the outer-loop algorithm searching for the optimal signal-timing settings. Most of these solution approaches remain only research tools without practical use because of computational intractability. This research proposes an efficient solution algorithm to the problem. An adaptive traffic signal control model is embedded in a simulation-based DTA model. For each inbound approach at an intersection of interest, the adaptive model uses upstream information and a dynamic rolling-horizon approach to project traffic flow conditions for a dynamic but short (projection) period. The adaptive model provides the signal settings during the entire process of traffic flow simulation and for every iteration of the solution algorithm. Thus, during the entire solution process, the experienced travel times and resulting traffic assignment flows are based on the adaptive (demand-responsive) signal settings, allowing the DTA flows and the adaptive signal settings to be generated simultaneously in a single-loop algorithmic structure. Simulation experiments illustrate the capabilities of the proposed approach.

AB - Dynamic traffic assignment (DTA) applications require traffic signal control data that are typically difficult to obtain and cumbersome to code in the required format. In addition, the evaluation of future scenarios requires future traffic signal settings consistent with the forecast demand. These signal settings are not known a priori and are costly to estimate. Intuitively, future signal timings need to be reasonably optimized so as to represent what the traffic management agency will do. In the literature, integration between traffic control and DTA models has been formulated as a bi-level or single-level optimization problem with system or user optimal constraints. Most existing solution procedures require certain nested structure with an inner-loop algorithm solving the problem of user-equilibrium or system-optimal assignment and the outer-loop algorithm searching for the optimal signal-timing settings. Most of these solution approaches remain only research tools without practical use because of computational intractability. This research proposes an efficient solution algorithm to the problem. An adaptive traffic signal control model is embedded in a simulation-based DTA model. For each inbound approach at an intersection of interest, the adaptive model uses upstream information and a dynamic rolling-horizon approach to project traffic flow conditions for a dynamic but short (projection) period. The adaptive model provides the signal settings during the entire process of traffic flow simulation and for every iteration of the solution algorithm. Thus, during the entire solution process, the experienced travel times and resulting traffic assignment flows are based on the adaptive (demand-responsive) signal settings, allowing the DTA flows and the adaptive signal settings to be generated simultaneously in a single-loop algorithmic structure. Simulation experiments illustrate the capabilities of the proposed approach.

UR - http://www.scopus.com/inward/record.url?scp=84857855671&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84857855671&partnerID=8YFLogxK

U2 - 10.3141/2263-12

DO - 10.3141/2263-12

M3 - Article

AN - SCOPUS:84857855671

SP - 103

EP - 112

JO - Transportation Research Record

JF - Transportation Research Record

SN - 0361-1981

IS - 2263

ER -