The Regional Dynamic Traffic Congestion Warning And Active Control Decisions

With the rapid development of automobile industry and the increasing level of urbanization, urban congestion problem becomes more and more serious. Traffic congestion has become an important factor of restricting the city development and impeding the travel of urban residents. On the one hand, the basic work of congestion management includes quickly identifing and efficiently predicting the traffic state and dynamically warning the traffic congestion. Essentially, congestion dynamic warning is a decision-making behavior, in fact, which includes rapid identification of traffic congestion and accurate estimate of traffic congestion trends. Congestion warning would provide the effective decision support for the traffic congestion control and describe the dynamic development of traffic state rather than describe congestion appearance. On the other hand, active congestion control decision refers to evacuating traffic congestion rapidly at the beginning, and avoiding further deterioration of congestion. Transport system has the characteristics of randomness, time-varying and nonlinear, therefore effective congestion control policy must be real-time in line with the traffic congestion situation and dissipation process. Traffic congestion control model should have the abilities of dynamic adaptation and active control.Main points of this paper are as follows:Firstly, the paper distinguished traffic state quickly. Due to the diversity of means and methods of traffic information collection, the basic information had the characteristics of multi-source and heterogeneous. In this paper, multi-source heterogeneous information was fused to determine real-time traffic status, in order to overcome the problems of information defect and data distortion owing to a single source of information. Basis for the characteristics of fuzziness and non-stationary of multi-source traffic information, The intuitionistic fuzzy set theory was introduced to set up the consistent intuitionistic fuzzy information fusion algorithm, and then traffic state decision information fusion model was established for real-time traffic state discrimination. This paper creatively put forward a new method for constructing intuitionistic fuzzy numbers based on dual membership functions, and then used the membership degree and the non-membership degree to build the support functions for getting decision-making information consistency measure model. Meanwhile, decision-making information integrated weights only depended on the support degree, the weights were updated dynamicly along with the change of decision-making information.Secondly, the paper predicted traffic flow parameters. This paper researched the method for predicting traffic flow parameters, including queue length forecasting at intersection and short-term traffic flow forecasting, under the environment of multi-source traffic information. On the one hand, queue length was an important decision variable of traffic signal control. In this paper, the initial queue length was determined based on the real-time data of floating car, using traffic wave theory. Then through circular computations, the accuracy of prediction was enhanced. This model overcame the problem that the initial queue length was difficult to determine. It was shown that the algorithm has strong adaptability of timing control, inductive control, intelligent control. On the other hand, short-term traffic flow prediction was to judge the future development of traffic flow based on the real-time data. In order to realize the traffic flow process prediction, this paper refined and mined effective information of traffic flow on road network, which reflected the traffic flow movement characteristics of space and time, established time-space model of traffic flow. The model included four submodels, they were the description of the road network model, the traffic distribution description model, the signal control description model and the inflow and outflow calculation method model, these models intuitionisticly described traffic flow operation and evolution mechanism.Thirdly, the paper set up the dynamic warning system for regional traffic congestion. In order to adapt to the demand of active management, the dynamic warning system needed to break through the static description of traffic state and reflect the crowded dynamic development trend. This paper put forward a new method of traffic dynamic measure, used traffic running reliability of the network to analyse the traffic status based on the precondition that the space headway had the different probability distribution under the different traffic state, determined the running reliability and built the traffic congestion evaluation model of the congestion region, quickly identified crowded areas, crowded regional boundary, crowded sections,and then made the alarming strategy.Fourthly, the paper made active decision system for traffic congestion based on signal control. This paper built the knowledge representation and reasoning methods to establish an open and adaptive congestion control model comprising a variety of decision variables and including diverse control rules. The adaptive weighted fuzzy petri net modeling method is put forward by improving the weighted fuzzy petri net, in which model the connection weights would be dynamic updated with the change of decision variable values, as result, the expression ability and accuracy was improved without increasing the scale of petri net. orchestrating the upstream and downstream traffic flow running state, building knowledge representation and reasoning model of traffic control and guidance strategy, the right of way in each phase was optimally distributed for guiding and controlling the saturated traffic flow quickly and effectively. What's more, in order to facilitate the model's implementation by programming, the matrix formal reasoning algorithm was designed.