Study On The Partition Methods Of Heterogeneous Network Oriented To Macroscopic Fundamental Diagram

With the advancement of urbanization and the rapid increase in the number of cars,traffic congestion on urban roads presents regional characteristics.This trend has increases the requirement of analyzing and grasping the urban traffic state.In recent years,the research found that macroscopic fundamental diagram of road network is an effective tool to reflect the state of traffic flow.And the homogeneity degree of road network is the key factor influencing the accurate analysis of MFD.In order to obtain the network sub-area with high homogeneity,it has become an important means to realize the rational division of road network by clustering method.The clustering algorithm brings about the ideas of improving the road network homogeneity.At the same time,it also faces the complex topological relationship of traffic network and the challenge of the distribution of actual traffic load.The workload and achievements are mainly reflected in two aspects.One is to use the measured data to evaluate the road network segmentation effect of the normalized segmentation algorithm in the literature.It is found that the segmentation result of the algorithm depends on the similar weights among the road segments.A penalty coefficient is added to the weight function between road segments,so that the segmentation algorithm can divide different results for different traffic flows.The second is to solve the shortcomings of classical K-means clustering algorithm in the application of heterogeneous road network partitioning.The road network connectivity is integrated into the algorithm to solve the problem of discontinuous clustering results in the application of road network partition.Use the maximum and minimum distance algorithm to determine the initial cluster center and the difference between the road segments.The dynamic frequency statistics of the road network partition at successive time intervals is used to merge and split unstable "noise" road segments to improve the stability and compactness of the road network in the sub-regions.One of the characteristics of urban construction in China is multi-center and multi-group.The corresponding use of road network presents the characteristics of multiple regions,which determines that traffic management needs to divide road network according to traffic characteristics.The road network partitioning algorithm proposed in this thesis has important theoretical and practical value.These two algorithms provide an important guarantee for the creation of well-performed macroscopic fundamental diagrams.