Evaluation of traffic signal timing policies for multiple objectives

Thumbnail Image
Date
2021-05
Authors
Yang, Yifan
Major Professor
Advisor
Day, Christopher M
Wang, Kejin
Wood, Jonathan
Committee Member
Journal Title
Journal ISSN
Volume Title
Publisher
Authors
Research Projects
Organizational Units
Journal Issue
Is Version Of
Versions
Series
Department
Civil, Construction, and Environmental Engineering
Abstract
Traffic signals are a critical element of roadway infrastructure, and policies that govern how they operate have an important impact on the performance of roadways where they are used. The determination of effective signal policies in accordance with differing traffic demands and conditions is a challenge, particularly because there are often competing objectives in the establishment of signal control. This study examines the effects of signal policies using a framework to evaluate their impacts from a multi-objective perspective, using the concept of the Pareto Front. The Pareto Front describes the set of optimal outcomes in a space defined by multiple objectives. By viewing the performance of a signal system from this perspective, a spectrum of possible outcomes can be observed resulting from changes in the control policies. This concept is applied to two case studies: an isolated signalized intersection and a signalized corridor, using results from simulation models. In both case studies, a tradeoff in the performance across different objectives is identified and documented using the Pareto Front concept. In the isolated intersection case study, policies on the design of detection zones are compared, showing a comparison between long and short detection zones. In the corridor case study, an array of signal timing plans are tested that include several modes of conventional operation (both coordinated and non-coordinated under a variety of options), as well as two adaptive control methods. In isolated intersection simulation, long detection zone scenarios without passage time and short detection zone scenarios with passage time are compared. Results show that long detection zone scenarios have lower delay than short detection zones. Moreover, variation of the maximum allowable headway reveals a Pareto front when considering a tradeoff between major and minor street delay. In corridor simulation, a Pareto front was also found in the delay of major and minor movements when considering alternative methods of corridor signal operation. A similar tradeoff is found when alternative performance measures are used (arrival on green and number of split failures). This Pareto front forms a useful reference for comparing the performance of real-time adaptive control algorithms. In this study, two algorithms were studied: a self-organizing algorithm and a schedule-based algorithm. The self-organizing algorithm was found to extend the feasible performance beyond the existing methods, while the schedule-based algorithm had performance similar to the existing conventional methods.
Comments
Description
Keywords
Citation
DOI
Source
Subject Categories
Copyright