Historical Travel Time Based Bus-Arrival-Time Prediction Model

摘要

Bus-arrival-time information service is a key component of an Advanced Public Transportation System (APTS). Instantaneous and accurate prediction of bus-arrival-time can improve the quality of transit service. With the input of tested historical bus travel time, a combined bus-arrival-time prediction model is promoted with a self adaptive exponential smoothing algorithm to predict interzone-section travel time and a statistical algorithm for multi-section travel time prediction. An experiment was conducted to compare the performance between the multi-section travel time (MSTT) based model and the sum of multi section travel time (STT) based model, with measures of mean error (ME), mean absolute error (MAE), and mean absolute percentage error (MAPE). The prediction results show that the MSTT based model outperformed the historical multi STT based model in terms of accuracy, reliability and precision.