Determination of Progression Speeds for Traffic Signal Coordination
Loading...
Authors
YUE, RUI
Issue Date
2020
Type
Dissertation
Language
Keywords
Machine Learning , Posted Speed Limit , Progression Speed , Signal Coordination , TranSync , VISSIM
Alternative Title
Abstract
Signal coordination has been widely implemented throughout the world as an effective approach to mitigating traffic congestion, improving operational efficiency on urban arterials, and reducing fuel consumption and emissions. Therefore, developing and implementing efficient signal timing plans is of significant importance. In the U.S., there are around 323,000 traffic signals currently operating in the nation’s roadway system. According to the signal timing manual, more than half of them need signal timing adjustments, and there is still a need for improvements in signal coordination plans. Inadequately implemented signal coordination plans may result in unexpected traffic delays, congestion, and even safety issues. Therefore, it has been a crucial task for transportation agencies to properly implement signal coordination plans. Multiple factors may influence the effects of signal coordination. Among these factors, progression speed is one of the most important, but its effects on signal coordination have not been thoroughly investigated. In this regard, this dissertation aims to develop a systematic approach to determine the optimal progression speeds under different conditions during the timing development stage. A major task of this research involved the development of an analytical model for determining progression speed. This model served as a foundation for developing a practical guide, as well as conducting field and simulation analyses. To achieve maximum efficiency, it is necessary to associate the progression speed and the traffic flow travel speed, which can be directly obtained through field studies. A significant amount of field data were collected and employed to derive the relationship between the average speed (an estimation to the speed of the platoon, which is also proximate to the optimal progression speed), and its influencing parameters. Three machine learning methods (Scikit-learn, TensorFlow, and Genetic Programming), and a traditional method (Linear Regression), were used to develop the models. Independent parameters that may contribute to average speed were considered, including segment spacing, grade, speed limit, volume, number of lanes, vehicle status at the upstream signal, and vehicle status at the downstream signal. The relationships between average speed and independent parameters were established. The results of the theoretical model are very sensitive when they are used in the field. To produce the results that can be applied in practice, a series of microscopic simulations were designed in VISSIM, a widely used commercial software package in the field of transportation engineering. Various scenarios were carefully designed before conducting the simulation experiment to ensure valid results . A sensitivity test was conducted to capture the impacts of various factors on traffic performance. Data from four signalized arterials in Reno, Nevada were used for the VISSIM simulation. The results of the study showed that the proposed models accurately reflected the relationship between vehicle travel speed and related influencing factors. A general conclusion was that when the traffic volume was low, e.g., volume to capacity ratio less than 0.7, using a progression speed slightly higher than the posted speed limit yielded better performance. When the volume was high, e.g., volume to capacity ratio larger than 0.7, a progression speed set at the posted speed limit was adequate. Deploying a progression speed lower than the posted speed limit was not advised. However, deploying a progression speed higher than the posted speed limit was not allowed since it violated the safety constraint. To avoid safety concerns, the progression speed should be equal to the posted speed limit. As a result, the posted speed limit is recommended as the progression speed as long as the real platoon speed is not significantly different than the posted speed limit, if the traffic platoon speed largely deviates from the posted speed limit, the real platoon speed should be considered to be used. Finally, the recommended progression speed guidance was tested in a real-world setting. New signal coordination plans were generated using the posted speed limit as the progression speed versus before timing generated by the investigated speed. Traffic performance data were collected before and after the implementation of the new signal coordination plans. The before-after comparison of results indicated that the new timing plans had significant improvements over the previous ones, further proving the validity of the proposed progression speed guide for practical applications.