Research On The Dynamic Partitioning Of Urban Road Network Control Sub-region

With the rapid expansion of the urban scale,the research of the evolution of traffic congestion has attracted widespread attention.The study of the congestion status of a single link has been unable to adapt to the characteristics of congestion spreading and dissipating on adjacent links.Therefore,the heterogeneous road network can be effectively divided into control sub-regions with different levels of congestion,which can visually obtain real-time congestion scenarios.Against this background,this paper studies the application of road network control subregion partitioning based on slow coherency theory,Dirichlet integral solution,and graph cutting theory to dynamic changes in traffic.The macroscopic fundamental diagram of sub-region is also studied.The main work of the paper is as follows:Aiming at the shortcomings of using the similarity characteristics between adjacent links to partitioning the control sub-region,a method of partitioning the control sub-region of the urban network considering the characteristics of congestion propagation was studied.Firstly,based on the improved depth-first search algorithm,the fast optimization of the associated path is achieved,and the calculation of similarity between non-adjacent links is solved.Then,based on the slow coherency theory,the good macroscopic fundamental diagram(MFD)of the control sub-region is obtained.Finally,the validity of the algorithm is verified by using the network data of the main urban area of Yiwu.Aiming at the shortcomings of the traditional static partitioning algorithm in the application of dynamic traffic changes,a dynamic network partitioning method based on the Dirichlet problem was proposed.Firstly,the unallocated links are reasonably divided by using the Dirichlet problem.Then,the partition result of the stable link is stored,and the link with low homogeneity is repartitioned.Finally,the road network partition in Farmers Branch,USA was used as an example,it is shown that dynamic partition is more efficient than static partition which maintains the original partition result.In order to solve the problem of subdivision in the case of classic subdivision dynamic partitioning algorithm that does not consider traffic flow mutation,a traffic control subdivision dynamic partitioning model is researched.Firstly,a correlation framework was built based on the Moran’s I index.Secondly,the slow coherency theory combining with graph cutting(referred to as ?-SC-FL)is used to establish a framework for network space partitioning.Based on the original partitioning results,the links with high heterogeneity in the sub-regions are identified by heuristic algorithm,and then they are iteratively fine-tuned,merged and divided to realize the dynamic partition of sub-regions.Taking the dataset of Farmer Branch in the United States as an example,the proposed algorithm has better partition effect than the classic partition algorithm.