Development of dynamic real-time integration of transit signal priority in coordinated traffic signal control system using genetic algorithms and artificial neural networks

Many transit agencies are interested in employing transit signal priority (TSP) at signalized intersections in urban areas to reduce travel time and achieve other operational benefits. However, state-of-the-art systems do not consider the influence of bus dwelling activities at bus stops, especially when the stops are located near the intersection stop line. They are also incapable of efficiently considering simultaneous priority requests and implementing TSP within coordinated traffic networks. This research develops a dynamic real-time control logic considering the influence of transit dwelling activities on priority requests and surrounding traffic performance. The problem of accurately predicting transit travel time in situations in which dwelling activities may take place before buses reach the stop line is first solved through the development of an Artificial Neural Network prediction model. The optimization of the prioritized and coordinated network traffic signals is then performed by using Genetic Algorithms. The resulting signal controller is tested in a simulated traffic network using the VISSIM microscopic traffic simulation software. The simulation results show that the proposed signal control model has the ability to improve transit and non-transit traffic operational performance. Test results more specifically show an ability to reduce delays and the number of stops incurred by transit vehicles, while improving schedule adherence and minimizing negative impacts on traffic resulting from the provision of transit priority treatments.