Traffic Congestion Discrimination And Prediction Based On Spatio-temporal Correlation Analysis

With the popularization of smart phones and taxi sof.tware,the development of mobile communication and taxi car GPS equipment,the relevant departments have accumulated a large number of floating car data(FCD)with extremely large levels.The FCD is based on the real driving path of the vehicle,and has the characteristics of wide coverage,good tracking,and complete trajectory that are not available in conventional traffic data.How to extract valuable information from such a large amount of data,provide guidance and basis for traffic management and control decision-making,and become the key demand of current traffic big data utilization.Based on the measured FCD,the study of the spatiotemporal characteristics of road traffic state characteristics,grasping and predicting the road traffic state is not only a new idea to improve the intelligence of intelligent traffic,improve its operational efficiency and data utilization.It also provides decision-making basis for improving road traffic congestion and improving traffic infrastructure and congestion mitigation strategies.This paper takes the urban road network traffic status as the research object,based on the measured FCD,combined with the advantages and characteristics of the data itself,and studies the traffic state between the upstream and downstream and the global road network in view of the interaction and constraints between urban road traffic states.Spatiotemporal characteristics,as well as the evolution of congestion.Based on this,the road traffic status is judged and predicted.The main research contents are as follows:1.Determine the quantitative indicators of spatiotemporal correlation research based on FCD.According to the road traffic flow parameters extracted from the FCD,the actual demand of the road traffic flow state and the road traffic state spatio-temporal correlation analysis cannot be fully described for the single evaluation index.Based on the traffic state threshold judgment of the FCM method,the structure is constructed.The spatiotemporal congestion index is used as the basic indicator for the analysis of the temporal and spatial correlation of traffic conditions.2.The spatial and temporal evolution of traffic congestion based on correlation analysis is studied.According to the separation and extraction of road traffic flow trend and detail vector,the existing Pearson correlation analysis is improved,and the generalized least squares method is introduced to make it more close to the road traffic state evolution.Based on the characteristics of traffic state classification,Kendall coefficient analysis is introduced to study the evolution of road traffic flow state.A high correlation path search method based on fusion coefficient is proposed as a research area determination method for traffic state prediction model.3.Analyze the temporal and spatial evolution characteristics of urban road traffic status.The road traffic state is extended from the simple upstream and downstream to the global road network.Based on the spatiotemporal Moran index,using the local Moran index to show the traffic congestion aggregation and diffusion characteristics,adding the Moran quadrant as the eigenvector,improving the random forest algorithm,and constructing a mixed forest model considering the temporal and spatial correlation of the road network traffic state,the road traffic The flow state is predicted in a short time.And verify by example.In summary,this paper proposes a method for judging and predicting urban road traffic state based on spatio-temporal correlation analysis.At the same time,it analyzes the evolution process of road network traffic state in the upstream and downstream connections and the global road network.The effectiveness of the above work and the sustainability of further research can provide effective support and reference for urban road traffic management and control.