A Reliable Hybrid Prediction Model for Real-time Travel Time Prediction with Widely Spaced Detectors

This paper presents a travel time prediction model that employs a small number of traffic detectors to perform real-time prediction under recurrent traffic conditions. The proposed model that consists of mainly a multi-topology Neural Network model and a supplemental component of an enhanced k-Nearest Neighbor model is capable of using various types of available information and contending with the potential detection errors and missing data. The evaluation results from field data have indicated that the developed hybrid model is capable of generating reliable prediction of travel times under various types of traffic conditions, and offers the potential for its application in a large freeway network.