| With the increasing expansion of city scale and the rapid growth of traffic demand,urban traffic congestion becomes more and more complicated,and regional traffic congestion has become a new normal in large and medium-sized cities of China.Regional-level traffic congestion problem causes various travel difficulties,which seriously restricts the development of urban economy and the improvement of residents’quality of life,and puts forward higher requirements for the level of traffic management.Based on in-depth analysis of urban traffic congestion causes,exploring regional-level congestion identification and optimal control methods for urban road networks has become one of the hot and key topics in the field of traffic control.Dynamic identification of congested regions is the key and foundation of traffic congestion control.It must dynamically reflect evolution of congestion and perceive development of congestion in advance,rather than simply describe appearance of congestion.Effective regional congestion control decisions must be consistent with dynamic congestion identification,which is to break through the hysteretic nature of conventional congestion control and to improve initiative and global optimization capabilities of traffic congestion control.Based on this,the dissertation studies dynamic identification methods and perimeter control strategies of traffic congestion regions,thereby providing theoretical supports for congestion management practices in large and medium-sized cities.Faced with the problem of urban regional traffic congestion,the dissertation first proposes a traffic state parameter extraction and credible fusion method by analyzing the characteristics of urban multi-source traffic data.Based on the evolution mechanism of traffic congestion,a dynamic identification method of congestion warning regions based on community detection is proposed to provide a necessary foundation for the study of active traffic congestion control decisions.Furthermore,a dynamic perimeter control strategy of single-region urban road network and a coordinated perimeter control strategy of multi-region urban road network are proposed by introducing the theory of macroscopic fundamental diagrams,which solves the problems of regional traffic flow optimization control and road network dynamic equilibrium under different congestion patterns.The main research works and achievements are summarized as follows:(1)Urban traffic state parameter extraction and credible fusion based on multi-source data.Utilizing the characteristic analysis of single-source traffic data,link travel time series are extracted from license plate recognition data,geomagnetic detector data and floating car data,respectively.Combined with the distribution fitting functions of link travel time series,a support degree algorithm based on similarity function and a credibility algorithm based on membership function are developed,aiming to overcome the conflicts among multi-source traffic data and the uncertainties of single-source traffic data.Furthermore,a link travel time fusion method based on credibility of evidence is proposed to obtain accurate and reliable traffic state information by integrating support degree with credibility.This research can provide necessary data support for the in-depth study of dynamic identification methods and perimeter control strategies of urban congested regions(2)Dynamic identification of traffic congestion warning regions based on community detection.On the basis of road network structure,link direction and time-varying traffic load,a dynamic topology model of urban road network is built.And automatic recognition of link congestion patterns is realized by the calculation of link weights.Combined with spatial and temporal evolutions of urban traffic congestion,a dynamic identification method of congestion warning regions based on community detection is proposed.This method includes three consecutive algorithms:discovering initial community,expanding community,and merging communities,which obtains congestion warning regions with spatial compactness and homogeneous states from heterogeneous urban road networks.This research can provide necessary theoretical basis for the study of single-region dynamic perimeter control and multi-region coordinated perimeter control based on macroscopic fundamental diagrams(3)Dynamic perimeter control of single-region urban road network based on macroscopic fundamental diagrams.For single-region traffic congestion in urban road network,the theory of macroscopic fundamental diagrams is used for macroscopic modeling of regional traffic flows,in order to actively evaluate network traffic dynamics.Based on the analysis of queue spillbacks at border links,dynamic border gated points are identified.Combined with real-time boundary conditions,a regional perimeter control strategy based on real-time traffic flows and queuing dynamics is proposed to adjust signal timing parameters of traffic inflows on the periphery of region.This can actively guide the development of regional traffic flows in a positive direction,reduce the possibility of queue spillbacks at border links and improve the overall operating efficiency of regional traffic flows.This research can provide beneficial reference for the study of coordinated perimeter control strategy of multi-region urban road network.(4)Coordinated perimeter control of multi-region urban road network based on macroscopic fundamental diagrams.For heterogeneous congestion in urban road network,a two-layer partitioning method is adopted to divide heterogeneous road network into multiple homogenous congested sub-regions and boundary sub-regions.And a traffic flow equilibrium model of multi-region network is constructed based on macroscopic fundamental diagrams,in order to accurately describe traffic flow operating dynamics of heterogeneous network.On this basis,a multi-agent based hierarchical traffic management scheme is designed,and then a transfer flow optimization model among multiple congested sub-regions is proposed.Furthermore,a multi-region coordinated perimeter control strategy using model predictive control is developed to alleviate multi-region traffic congestion and to achieve global optimization of traffic flows in urban road network.This research can provide effective theoretical supports for the establishment of traffic congestion control system of large-scale urban road network. |
