An Application of Neural Network in Corridor Travel Time Prediction in the Presence of Traffic Incidents

Incident related traffic congestion leads to enormous economic loss each year in the world. To predict the traffic situation when an incident occurs and disseminate the information to the traveling public can alleviate the traffic congestion caused by the incident. This research collected traffic incident data and traffic condition data from a highway corridor (Interstate Highway 66 Eastbound) in Northern Virginia. After pre-processing, data fusion between the two different data sources was successfully conducted and the cross reference between the incident and traffic condition data sets was set up. Based on the fused data sets, a neural network model for corridor travel time prediction in the presence of traffic incidents was developed. After the model was trained and optimized, randomly selected new data was used to test the performances of the proposed model under three different scenarios: (1) the input variables were incident related information only; (2) the input variables were current traffic condition information only; (3) the input variables included both the incident related information and the current traffic condition information. The performance indicators of the model were calculated under the three scenarios, and the statistics were compared. The results demonstrate that it is possible to accurately predict the future travel time within a corridor in the presence of traffic incidents when given sufficient amount of data. With exceptional learning ability, neural network is proven to be an effective tool in modeling this travel time prediction problem. The developed neural network delivers a good fit in most cases, indicating that it is a successful model. It is also observed that incident related information roughly dictates the trend of the impact on traffic, while current traffic condition provides a dynamic environment where the incident occurs. Addition of current traffic condition information can further improve the prediction accuracy. The predicted travel time information is valuable for traveler information systems and traffic incident management systems.