A Study Of Urban Trunk Road Traffic Condition Classification And Recognition System Based On Small-sized Probe Reports

Probe vehicle system is an advanced technology of traffic information collection,and has come to the stage of commercial application. But in the early stage, the numberof probe vehicles is so small that sample size is inadequate and accuracy of trafficidentification is not high. However, it is an important stage to get users to trust thesystem, so that, under the case that there is little probe vehicles, an effective trafficclassification is particularly important.The core content of this article is, under the condition of little probe vehicles,building traffic condition classification, designing recognition system, and verifying itusing simulation.In this article, the traffic condition is divided into three phases: free flow, crowded,congested, by the percentage of vehicles departing at the downstream in the same greentime in vehicles entering the same link at the upstream in the same green time. First, weselect Link travel time, average speed of upstream section, and link waiting time as theparameters of the single car, and study the distribution characteristics of threeparameters. Then, clustering algorithm and the minimum risk-based theory of Bayesianalgorithm are separately used to identify traffic condition. The minimum risk-basedtheory of Bayesian algorithm has a higher recognition rate, so that it is used to designrecognition system. Finally, on the microscopic traffic simulation platform VISSIM, weestablish road network, verify the recognition system using the data that we get from thesimulation. High recognition rate and low error rate are reached.Under the condition of small-sized probe report, the recognition system oftraffic condition designed in this paper would recognize different trafficconditions with a high accuracy, and solve the crisis of confidence issues in theearly stage of building the probe vehicle system. This thesis is highlyapplicable.