Dynamic Dilemma Zone Modeling and its Protection.
详细信息
文摘
As a major cause of crashes at high speed signalized intersections, the issue of dynamic yellow light dilemma zone DZ) has been raised by researchers for many years. However, quantitative study of the inherent factors contributing to the DZ dynamics remains an issue, especially for Type I dilemma zone and option zone. In this context, this research aims at quantitatively modeling the dynamic nature of DZ through investigating contributing factors to Type I and Type II dilemma zones, respectively. To fulfill this goal, video data collection has been conducted at four high speed signalized intersections in Ohio. Qualified trajectory data of 1445 vehicles have then been extracted from 46-hour digital videos using customized software program VEVID. Results of statistical analyses of the vehicle trajectory data have quantitatively revealed that the drivers minimum perception-reaction time PRT) is a function of speed, while the maximum deceleration and acceleration rates are functions of both speed and the aggregated 85th percentile speed of the intersection approach. On the other hand, results of binary logistic regression analyses have indicated that lane position, vehicle type, posted speed limit, duration of yellow interval, and speed gap between following vehicles are significant contributing factors that influence drivers stopping probability, in addition to vehicles speed and location at the onset of yellow indication. These findings have successfully revealed and modeled the dynamics of Type-I and Type II DZs, and have provided theoretical basis for updating the existing DZ models to reflect DZs dynamic characteristics. As a solution to safety issues caused by DZ, the optimal advance detector placement method for DZ protection has been specifically developed based on the updated dynamic DZ model resulted from this research. The optimization process is conducted in a calibrated microscopic traffic simulation test bed. The optimization objective is to minimize the combined cost of the traffic conflicts due to the existence of DZ, which represent the safety, and the delay experienced by drivers, which represents the mobility. Specifically, the probability of traffic conflicts is assessed by dilemma conflict potential DCP), which is a comprehensive surrogate dilemma hazard model proposed in this research. Unlike the traditional surrogate dilemma hazard measure of "number of vehicles in DZ", DCP quantitatively computes the combined probability of rear-end and right-angle conflicts faced by each vehicle that approaches the intersection during the yellow interval. The computation is based on both the speed and the location information of a target vehicle and its leading vehicle with regard to the locations of DZs as well. The DCP model has been calibrated by using field-observed trajectory data before its use in the optimization process. Finally, the generated optimal advance detector placement method has been evaluated with comparison to other four classic placement methods which are widely used in the US. The evaluation results have proved that the optimal method is superior to Bonneson, Beirele, SSITE, and Winston-Salem layouts by greatly reducing the combined cost of dilemma hazard and delay, which indicates it can provide effective and efficient protection to dynamic DZ.