AUTOMATED DETECTION OF LANE-BLOCKING FREEWAY INCIDENTS USING ARTIFICIAL NEURAL NETWORKS

A major source of urban freeway delay in the U.S. is non-recurring con gestion caused by incidents. The automated detection of incidents is a n important function of a freeway traffic management center. A number of incident detection algorithms, using inductive loop data as input, have been developed over the past several decades, and a few of them a re being deployed at urban freeway systems in major cities. These algo rithms have shown varying degrees of success in their detection perfor mance. In this paper, we present a new incident detection technique ba sed on artificial neural networks (ANNs). Three types of neural networ k models, namely the multi-layer feedforward (MLF), the self-organizin g feature map (SOFM) and adaptive resonance theory 2 (ART2), were deve loped to classify traffic surveillance data obtained from loop detecto rs, with the objective of using the classified output to detect lane-b locking freeway incidents. The models were developed with simulation d ata from a study site and tested with both simulation and field data a t the same site. The MLF was found to have the highest potential, amon g the three ANNs, to achieve a better incident detection performance. The MLF was also tested with limited field data collected from three o ther freeway locations to explore its transferability. Our results and analyzes with data from the study site as well as the three test site s have shown that the MLF consistently detected most of the lane-block ing incidents and typically gave a false alarm rate lower than the Cal ifornia, McMaster and Minnesota algorithms currently in use.