Identification Method Study On Traffic Flow Evolvement Based On Statistical Change-point Theory

Traffic flow evolvement is usually ignored in traditional research methods of traffic flow states identification,which are restricted by mathematical model and mainly focused on identification in advance(traffic flow prediction)and real-time identification(incident detection or traffic flow qualitative change detection).The evolution process of traffic flow is generally classified as quantitative change and qualitative one.It is important to study the traffic flow evolvement.In reality,if one road happened to be congested,its neighbor segments are most likely to be involved.The urban traffic congestion mechanism and the reason for local traffic paralysis can be explained essentially trough study on traffic flow evolvement of the crowded bottlenecks.According to the evolvement,traffic vehicles can be induced in time and traffic accidents can also be avoided.On the basis of the common intelligent algorithms and gamma distribution change-point test algorithm,the issues studied in this paper are listed as follows.Three Change-point analysis methods of traffic flow are put forward with probability change-point model and mean change-point model.A conclusion that distributions of free flow,congested flow and intermittent flow can all be replaced by gamma distribution is come up with.Traffic flow evolvement identification method is proposed based on gamma distribution change-point test algorithm.Considering nine different traffic flow evolvement,gamma distribution shape parameter adapting to traffic flow evolvement is estimated with maximum likely estimation method.In view of the vagueness and uncertainty of traffic flow,the optimal sampling interval of traffic flow is researched based on fuzzy comprehensive evaluation.Considering the randomness and uncertainty of traffic flow,dynamic bayesian network is used to identify the traffic flow states.The above study has significant meaning in easing traffic pressure in crowded areas,avoiding secondary accidents,saving energy and reducing pollution.