Bus Travel Time Prediction Using AVL and APC

Abstract

The focus of this paper is on using AVL (Automatic Vehicle Location) and APC (Automatic Passenger Counter) empirical data to develop a travel time model capable of providing real time information on bus arrival time to passengers (traveler information services) and to transit controllers for applying proactive control strategies. Three techniques have been used to develop bus travel time prediction models. The evaluation of these techniques showed that the time-lag recurrent neural network model outperformed other models in terms of accuracy and transparency demonstrating its ability to update the model based on new data that reflect the changing characteristics of the transit-operating environment. INTRODUCTION Many public transit operators have installed and used AVL (Automatic Vehicle Location) technologies to monitor in real time the locations of transit vehicles along transit routes. Several researchers have developed models to use such dynamic information for predicting bus arrival times at downstream bus stops (examples include Kalaputapu and Demetsky, 1995. Wall and Dailey, 1999; Lin and Zeng, 1999). The motivation for developing these models was mostly for providing information to transit riders on expected bus arrival times at bus stops. As such, the models included very simple independent variables such as historical link travel times, upstream schedule deviations, and headway distributions, in addition to the current location of the next bus. This study is part of a larger study that aims at developing an integrated system for dynamic operations control and real-time transit information. Currently, almost all transit operators implement control strategies, such as bus holding and expressing, after detecting schedule/headway deviations in the system, hence reactive in nature. The proposed system (shown in Figure 1) takes a proactive approach to operations control that would enable the controller to implement preventive strategies before the actual occurrence of deviations. This innovative approach requires the use of arrival time models sensitive to the considered control strategies. This paper focuses on developing models of such characteristics and on comparing the performance of various techniques for modeling bus arrival time models. Graduate Student, Department of Civil Engineering, University of Toronto, 35 St. George Street, Toronto, Ontario, Canada M5S 1A4 Tel: (416) 978-5049 Fax: (416) 978-5054, e-mail: farhan@ecf.utoronto.ca 2 Ph.D., P.Eng., Assistant Professor, Department of Civil Engineering, University of Toronto, 35 St, George Street, Toronto, Ontario, Canada M5S IA4, Tel: (416) 978-5907, Fax: (416) 9785054, e-mail: amer@ecf.utoronto.ca 3 Ph.D., P.Eng., Associate Professor, Department of Civil Engineering, University of British Columbia, 2324 Main Mall, Vancouver, British Columbia, Canada V6T 1Z4, Tel: (604) 8224379 Fax: (604) 822-6901, e-maih tsayed@civil.ubc.ca