Understanding how congestion at one location can cause ripples throughout large-scale

Understanding how congestion at one location can cause ripples throughout large-scale transportation network is vital for transportation researchers and practitioners to pinpoint traffic bottlenecks for congestion mitigation. of the proposed method. Results show that the prediction accuracy can achieve as high as 88% within less than 6 minutes when the model is implemented in a Graphic Processing Unit (GPU)-based parallel computing environment. The predicted congestion evolution patterns can be visualized temporally and spatially through a map-based platform to identify the vulnerable links for proactive congestion mitigation. Introduction Traffic congestion costs billions of dollars in each year due to lost time, wasted fuel, excessive air pollution, and reduced productivity. The 2012 Urban Mobility Report Ibudilast indicates that the annual average delay per person was 38 hours in 2011 for the 498 surveyed urban areas, which is equivalent Ibudilast to a 238% increase compared to that in 1982. Traffic congestion incurred a total of 5.5 billion hours of travel delays and 2.9 billion gallons of extra fuel consumption in 2009 2009, which corresponds to a congestion cost of 121 billion dollars [1]. Diagnosing congestion onset and predicting congestion evolution patterns Ibudilast are considered strategic countermeasures to locate traffic bottlenecks and adopt proactive measures for congestion mitigation. Many research efforts have been made to attain these goals [2, 3]. Nevertheless, the majority of previous studies have a tendency to view congestion areas within a small-scale network separately. As stated by Ibudilast Yang[4], the Ibudilast amount of network links for nearly Mouse monoclonal to NACC1 all of the existing visitors congestion prediction strategies do not go beyond 100. Furthermore, these scholarly research depend on either mathematical equations or simulation ways to depict the network congestion evolution. This often leads to suboptimality since transport activities involve individual factors that are challenging to represent or model accurately using mathematics-driven techniques. Prior network-wide congestion research mainly holiday resort to either complicated network theory [5C11] or visualization methods [12] to comprehend the advancement of network-wide visitors congestion. In complicated network theory, transport systems could be abstracted as scale-free systems [9], and visitors movement dynamics on the network are generated in line with the billed power rules distribution [11] Nevertheless, these assumptions aren’t adherent to the truth often, and lack enough visitors sensor data to validate their results. Visualization methods can intuitively screen the temporal and spatial distribution of network congestion by way of a map-based system, but are not capable of detailing the system of congestion era and predicting upcoming craze of congestion advancement. Within the last decades, tremendous visitors sensors have already been deployed on the prevailing freeway systems, producing plenty of data at about time resolutions relatively. The increasing option of network data can help you simultaneously examine visitors flows on the large-scale roadway network and take notice of the advancement of congestion on that network through data mining methods. Transportation network includes a large number of links with changing visitors condition as time passes. This is equal to a high-dimensional space where congestion prorogates and spatially temporally. This is complicated to model using traditional data mining techniques because of the curse of dimensionality: once the insight dimensionality increases, the mandatory training data grow [13] exponentially. The recent introduction of deep learning theory can address the curse of dimensionality concern through distributed representations, and thus retains great promise in learning high-dimensional features with huge data. Compared to those shallow learning architectures, deep learning is able to model complex non-linear phenomenon using distributed and hierarchical feature representation [14, 15], and has received numerous.

Comments are closed.