Research On System Optimum Dynamic Traffic Assingment Problems For Urban Network Based On CTM

Accurately and reasonably adjusting the distribution of traffic demand on the traffic network,and then achieving the dynamic equilibrium between supply and demand of road system is the core idea of urban active traffic management.There are some underlying problems in CTM based system optimum dynamic traffic assignment.For instance,some system optimum dynamic traffic assignment models can not be perfectly applied to the urban network with multiple OD pairs,and some models contain traffic holding-back,causing many traffic management technologies based on CTM can not be applied to large-scale urban network.Therefore,this thesis proposes a system optimum dynamic traffic assignment model with non-holding back flows for urban network based on path CTM.The research includes three parts:establishment of the path based CTM model for urban network,the traffic holding-back problem of SO-DTA at network level,and the modelling of SO-DTA with non-holding back.The details of this research are summarized as belows.The path based CTM is dynamic network loading process of SO-DTA,and is also one of important components of SO-DTA problem research.Starting from the basic principle and mathematical form of Daganzo’s CTM,this thesis uses path based CTM to build traffic flow model for urban network by considering the limitations of CTM,e.g.the turning propotions need to be presetted,and CTM can not be perfectly applied to urban network with multiple OD pairs.The path based CTM for urban road network improves the basic logic of CTM by adding the "path" dimension in the CTM variables,and the improved model can accurately handles the turning flows in diverging cell connectors and merging cell connectors according to upstream and downstream traffic states without presetting the turning propotions.In the research of traffic holding-back problem of SO-DTA at road network level,the mechanism of traffic holding-back is deeply explored at mathematics level,this thesis reveals that the linearized CTM loading in SO-DTA will cause the multi-value of flow modes and the uncertainty of traffic states,and lead to traffic holding-back phenomenon.A non-holding back linear programming formulation for road network with single OD pair(NLP)is proposed by considering the cause of traffic holding-back,and NLP model’s traffic holding-back elimination ability was tested and verified in the test netwok of Ziliaskopoulos.The path based CTM was taken as DNL model of SO-DTA,and a SO-DTA model with non-holding back flows for urban network was proposed by considering the route choice.Based on an actual urban network that includes 23 OD zones,18 roads and 31 intersections,SO-DTA model with non-holding back flows for urban network was established by LINGO 11.0 in the pre-setted simulation environment.The results,including time-varying probabilities of effective path set,cellular aggregate occupancy matrix,cellular aggregate flow matrix,cellular path occupancy matrix and cellular path flow matrix,were analyzed strictly to verify the model’s results do not contain traffic holding-back.The route choice probabilities of three selected OD pairs were estimated by using vehicle lisence plate recognition data of partial intersections of the research network,results show that vehicle’s route assignments of three OD pairs match well with route choice behaviours in actual scence.