Unlocking the Mysteries of Freeway Sensor Data to Diagnose Detailed Bottleneck Dynamics

摘要

The proliferation of fixed freeway sensor data has opened the door to more detailed analysis of freeway operations and capacity. The availability of more data (usually from fixed points on the network) also brings data management and processing challenges. Past research has often used arbitrary temporal aggregation as a means of smoothing data measured at single points. Comparisons are difficult to draw across segments and details of interest are often filtered out. This paper describes how the spatial and temporal evolution of traffic conditions on a freeway can be diagnosed using sensor data retained in their most raw form. The diagnostic tools used in this paper include curves of cumulative vehicle arrival number versus time and cumulative occupancy versus time constructed from data measured by neighboring freeway loop detectors. Once suitably transformed, these cumulative curves provide the measurement resolution necessary to observe the transitions from freely-flowing to queued conditions and to identify some notable, time-dependent traffic features surrounding freeway bottlenecks. Given that little is known about how traffic flows through bottlenecks, a greater understanding is required to formulate, to enhance or to verify mathematical models of vehicular traffic, so that they are consistent with the actual traffic features that are found to be reproducible. This understanding is also required before one can conclude whether or not bottleneck flows can be increased by eliminating or postponing freeway queues with control measures such as ramp metering.