A pattern matching approach to speed forecasting of traffic networks

[摘要] ObjectiveUnlike other existing traffic data collection techniques, probe vehicles, or floating cars traveling on a road network, have appeared as a complementary solution for increasing coverage areas without requiring expensive infrastructure investments. When organized in a fleet with communication capabilities and exchange of information with a central data system, they give rise to a Floating-Car Data (FCD) system. The purpose of this paper is to present a model for short-term traffic speed forecasting based on an operating FCD system, developed and operated in Italy, delivering real-time traffic speed information throughout the Italian motorway network and along some important arterial streets located in major Italian metropolitan areas.DesignSpecifically, a database covering the whole period ranging from April 2008 to October 2011 is available for Rome Ring Road, a toll-free motorway that encircles Rome (Italy), and the developed case study pertains to a portion of its available speed data.MethodA Pattern Matching method of prediction will be detailed, which reports interesting properties in terms of forecast accuracy; the method tries to identify, in the past data history, patterns neighboring (in a proper sense) the current one, which describes the actual traffic load, and produces forecasts supposing that the current situation will evolve in a similar way.

[发布日期] 2014-02-25

[关键词] Floating car data (FCD);Speed time series;Short-term speed forecasting;Pattern matching