Development of a performance prediction algorithm for evaluation of a traffic adaptive signal system for isolated high-speed intersections.

摘要

Drivers approaching isolated, high-speed intersections must make a decision to either proceed through the intersection or to stop when the phase changes from green to yellow. Each driver decides to proceed or stop based on his or her perception of the intersection. Differences in driver perception create an area called the dilemma zone, where the driver population demonstrates the greatest variation in decisions to proceed or stop. There are collision risks associated with either decision, so it is undesirable for vehicles to be in the dilemma zone at yellow onset.; Typically, some type of detection is placed in the dilemma zone so that the green phase is extended while a vehicle occupies the dilemma zone. This detection reduces collision risks, but it often results in longer green phases. A green phase that extends long enough can reach its maximum allowable time (max out) and end immediately even if vehicles are in the dilemma zone, creating the situation it was meant to protect.; A new type of dilemma zone protection system, the Detection-Control System (D-CS), was recently developed to reduce both collision risks and delay related to dilemma zone protection. D-CS has been tested in the laboratory and in the field. However, the scale of the testing was limited, and existing traffic signal models proved inadequate to describe D-CS' operations.; This dissertation develops a new performance prediction algorithm that provides output about D-CS performance. The performance algorithm determines the probability of reaching each possible end state within D-CS, the green phase duration for each end state, and the expected green phase duration covering all possible end states. The performance algorithm also provides the number of vehicles in the dilemma zone at yellow onset.; A simulation model is used to calibrate the performance algorithm to real-world conditions, and the calibrated performance algorithm is compared to field operation of D-CS. An investigation of the performance algorithm's predictions allows recommendations to be made for placing the D-CS detector traps, the protection zone width, maximum green, and the length of Stage 1 to help preserve both safety and operational efficiency at future D-CS controlled intersections.