Spatial And Temporal Correlation Analysis Of Urban Traffic Network At Multiple Levels

With the process of urbanization and social and economic development in China,the living standard of residents has increased significantly,and the number of motor vehicles in cities has increased dramatically,resulting in frequent road congestion.Therefore,by integrating multidisciplinary theoretical knowledge to study the traffic problems,we can grasp the spatio-temporal relationships and potential correlations in urban traffic networks,which can not only provide new ideas and scientific support for traffic congestion management,but also provide references for urban traffic planning.In this paper,the multi-level association mining and analysis of urban traffic network is taken as the main line of research,and the research is carried out at three levels: nodes,road sections,and road networks,respectively.In order to ensure the quality of the data used,this paper preprocesses the data in three aspects: anomalous data,missing data and data smoothing,and characterizes the preprocessed data in two dimensions: time and space,laying the foundation for the subsequent research.Firstly,the node correlation analysis is based on complex network theory,and the way of constructing complex networks of road networks is explained.The shortest path matrix is introduced to calculate the static characteristics of each node,and the node correlation degree is defined by combining the static characteristics with the dynamic congestion characteristics to quantify the strength of the correlation between each node and other nodes in the network.And the node correlation indexes of the Beijing expressway road network are calculated and analyzed,and the results show that the strongly correlated nodes are mainly distributed in the nodes connecting the inner ring area to the main urban area,and the correlation indexes of the nodes in the inner ring road are generally higher than those in the outer ring road.Then,the road section association analysis is carried out based on the improved Apriori algorithm.The concept of spatio-temporal association rules of road sections is proposed,and the improved spatio-temporal Apriori road section association rule mining algorithm considering the time is proposed from the perspective of data structure and algorithm logic by adding spatio-temporal constraints in the initial algorithm connection step,and the example analysis verifies that the improved algorithm can mine the association between congested road sections.Finally,in order to study the correlation between multiple road sections in the road network,compared with the traditional method,this paper combines Pearson correlation coefficient coefficient and DTW distance for similarity measure of traffic flow time series from the perspective of temporal features,introduces spatial reachability matrix from the perspective of spatial features,and carries out hierarchical clustering based on RP-DTW road section spatio-temporal similarity measure matrix.The experimental results show that the spatio-temporal clustering method can still excavate the highly correlated road sections in the road network well under the consideration of both temporal and spatial conditions,and analyze the change patterns and correlations of traffic flow parameters in the road network under different time periods,and its research results have certain guiding significance and practical application value for the large-scale traffic coordination control of urban traffic road networks.There are 44 figures,23 tables and 52 references in the main text.