Prediction Model of Bus Arrival Time at Signalized Intersection Using GPS Data

Abstract

Accurate prediction of bus arrival time at the stop line is a vital element to the bus signal priority system, but most previous approaches focused on predicting bus arrival times at next bus stops only. This paper develops a travel time prediction model to predict bus arrival time on the basis of global positioning system (GPS) data. Bus travel time from the detected location to stop line is divided into three parts: travel time from present point to the end of anterior queue, waiting time for the green light, and time for discharging anterior queue vehicles. Case studies were conducted in real-life signalized intersections to evaluate the performance of the model. Results showed that the presented model provided acceptable prediction accuracy. In addition, by considering the log interval of GPS data and prediction error, a determination method of the optimal decision-making zone when using sequential GPS data was developed.