摘要: |
The Oregon Department of Transportation (ODOT) has recently begun changing their standard traffic signal control systems from the 170 controller running the Wapiti W4IKS firmware to 2070 controllers operating the Northwest Signal Supply Corporations Voyage software. Concurrent with this change in standard signal control systems, ODOT has taken the opportunity to install test sites with adaptive signal control systems and evaluate advanced features in the Voyage software. The evaluation of advanced features and adaptive signal control systems has led to a series of questions about how to measure performance, when to apply a given feature and, when should one system be preferred over another. To answer these questions a survey of literature and practicing professionals was conducted to determine the current state of the practiceregarding conventional and adaptive signal control systems. The survey of practitioners indicated that practitioners in general were seeking answers regarding when and how to implement adaptive systems. Similar questions were found in literature, with the addition that the FHWAs Model Systems Engineering Documents for Adaptive Signal Control Technology added questions regarding whether existing systems had potential performance gains available through feature enablement. This knowledge was used to create an evaluation framework to guide practitioners in evaluating the performance of their current systems. A decision support framework based on decision tree logic and queuing theory models was built on top of the analytical framework to analyze the existing system for features that may improve performance. The decision support framework also provides a means of estimating the performance of different control strategies given the existing conditions. As a means of evaluating the various systems and selected features, the research team created a series of simulation models in VISSIM 5.30. These simulations were controlled via external logic emulating the signal control logic of the various systems and features. In total, 4,536 simulation cases were examined. The results of these simulations were used to calibrate the decision support and queuing model logic. |