Reverse Modeling Of Road Network Congestion Propagation

With increase in urbanization and consequent growth in vehicle ownership,urban traffic congestion is becoming more and more serious.As urban road network is a connected whole,even a small-scale congestion can propagate widely through roads.The study of traffic congestion propagation can help us understand the structure and function of urban road network,and then develop congestion control strategies.It is of great significance to alleviate traffic congestion,deal with emergency situations and ensure the smooth operation of urban traffic system.Although many scholars have used micro traffic flow model,macro traffic flow model and complex network dynamics model to study congestion propagation process,there is still a lack of a proper model that can reflect the actual traffic network congestion propagation process and its required data are easy to obtain.To solve this problem,reverse modeling theory of complex networks is used to model the congestion propagation dynamics of urban road network in this thesis.Reverse modeling theory is a new approach supported by data to studying network dynamics,which simplifies the assumptions and constraints in traditional modeling and can deduce network dynamics model from real data.Based on the introduction of reverse modeling theory,this thesis takes the road network of Chengdu as the research object and the taxi GPS trajectories as the main data to deduce the congestion propagation dynamics model.The main contents of this thesis are as follows:(1)The average speed of the corresponding road every 5 minutes is extracted from the taxi GPS trajectories data through GPS map-matching algorithm.For the road without GPS data at certain instant,its speed is considered as the average speed of adjacent roads.Then the average speed of all roads in 288 time periods in a whole day is obtained.(2)A directed dual network of road network is established in the thesis,and the road average speed is transformed into the congestion state of the nodes in the network by using the congestion state function.Based on the node states,a series of characteristics indexes are extracted by observing the network response to external perturbations.This thesis further puts these characteristics indexes into reverse modeling theory to push back the dynamics model,and obtains the congestion propagation dynamics model with parameters.For these parameters,the actual traffic data are used to estimate them step by step,and the complete congestion propagation dynamics model is obtained.(3)The congestion propagation dynamic model is used to analyze the actual traffic problems.Firstly,the average congestion state and the number of congested sections with time under different congestion degrees are studied.According to the simulation results of congestion propagation dynamics model,the propagation regular pattern of road network congestion is analyzed.Secondly,the roads are sorted based on the influence of congestion propagation,and the common characteristics of roads with greater influences in their own structure and surrounding facilities are analyzed.Finally,combined with the sorting results of degree,betweenness centrality and congestion propagation influence,the congestion mitigation effects of different congestion control strategies are compared.