Transit travel time prediction for passing signalized intersection using vehicle positioning data

This paper explores a practical method for predicting transit vehicle travel time when passing signalized intersections. Although the task provides important functions and serves many different uses for services providers and management, the issue has not been well solved due to the challenge of inferring delays at intersections. In this paper, the algorithm is investigated for calculating the delay caused by the existence of intersections, which is one of the steps required to predict transit arrival time. Our technic mainly includes: Firstly, decomposing the process of transit passing the traffic nodes. When delay occurs at intersection, the transit vehicle will experience four phases, including slowing down, waiting in the queue/service time, waiting for discharge and discharge. Secondly, the delay is defined as time differential between phase two and four and can be calculated and the arrival time from field test data is calculated to evaluate the proposed model. The proposed algorithm is meaningful to evaluate transit system and improve transit priority plan as well as presenting delay resulted from different causes.