01033613

Component

This paper describes how vehicle travel time forecasting is an important element for transit operations management and control. This paper develops the methodology used to estimate the vehicle operating parameters for predicting vehicle travel times and associated vehicle trajectories using primitive automatic vehicle location (AVL) data. The term “primitive” in this case is used to describe AVL data that simply includes the vehicle location and time stamp, without direct information on the vehicle’s proximity to a specific location (e.g. stop, station, intersection, or time point) on a route. Under three different assumptions of vehicle operating behavior, three methods are proposed to use AVL data to derive the vehicle travel times. Relatively, the assumption of correlation of travel times across days (a “trip-specific” assumption) is superior to the assumption of correlation between adjacent trips (a “day-specific” assumption), according to the regression results. A two-stage regression method is developed to integrate both the trip-specific and the day-specific assumptions, to yield a combined model. Regression results based on this combined model suggest that, with the parameters calibrated from the trip-specific model, the previous vehicle’s travel time prediction error has a significant impact on the travel time prediction for the current vehicle. Therefore, such a combined model is expected to improve the precision of the vehicle travel time prediction. Furthermore, a performance comparison of the trip-specific model and the combined model using a validation data set shows slight superiority of the combined model over the trip-specific model. However, the vehicle travel time prediction errors are still fairly large for both models. This may imply significant temporal variation of the vehicle travel times and operating parameters.

01020180

06-2372

English

19p

2006

Transportation Research Board 85th Annual Meeting

CD-ROM

Figures (4) ; References (15) ; Tables (7)

Automatic vehicle location; Information systems; Public transit; Regression analysis; Traffic forecasting; Transit operating agencies; Travel time; Trip purpose; Vehicle trajectories

Travel Time Prediction System

Data validation

Data and Information Technology; Highways; Operations and Traffic Management; Planning and Forecasting; Public Transportation; I72: Traffic and Transport Planning

Transportation Research Board Annual Meeting 2006 Paper #06-2372

TRIS, TRB

Mar 3 2006 10:58AM