Integrated ANN-Bayes-based travel time prediction modeling for signalized corridors with probe data acquisition paradigm

Abstract

Travel time is a major information in support of Transportation System Management and Operations (TSMO). Accurate travel time information extracted from probe data sources, which have been evolved in recently years, has led to the considerable research into travel time prediction modeling on freeways. However, little research has been reported on dealing with predicting signalized arterial travel time. It is mainly because of the natures of interrupted traffic flows along a signalized corridor or dynamic urban network, wherein varying signal timing schemes, crossing traffic, and dynamic route choices make arterial travel time estimations much more challenging than freeways. This paper presents an innovative methodological framework that can be used to integrate the exponential smoothing technique, Artificial Neural Network (ANN) techniques and Bayes algorithms for predicting travel time along a signalized corridor. The data source obtained along the US-27 corridor in the State of Ohio, was used to test two types of the developed models, i.e., hourly instantaneous travel time prediction model and realized (or experienced) route travel time prediction model. The study results indicate that the developed models can effectively capture the travel time patterns by combining an exponential smoothing base profile and an ANN-Bayes-trained dynamic profile. In this way, more sensitive changes can be identified to a variation aroused by non-recurring congestion. The testing results with real-world travel time datasets indicates a good performance of the predicted travel time models. The most errors range from −0.608 to 0.485, which are much lower than the threshold standard deviation if 2.24 and close to 6% of mean travel time value of 9.08 min. Overall, the study results suggest that the proposed methods are capable of capturing traffic patterns even when an incident occurs by comparing with conventional methods.