A Multi-Mode Traffic Congestion Prediction Method Of Urban Road Network Based On Pignistic Interval Length

In recent years,with the gradual progress of urbanization in China,the gradual increase in car ownership,imbalance of the relationship between traffic supply and demand is more and more serious.Traffic congestion has become an important factor affecting people's quality of life,at the same time,traffic congestion has become a bottleneck to curb the development of urban economy.because of the conspicuous problems like air pollution,noise pollution and other issues leading by traffic congestion.The assessment of road running status is the basis for studying the trend of road traffic status.In this regard,domestic and foreign scholars analyzed from the micro,medium view,macro and other different angles and made a series of results,but the research based on road network which is used for the study of the road traffic mode of operation identification and for the study of traffic congestion trends is relatively less.Based on the actual demand of urban traffic management,this paper starts from the point of view of data mining and pattern recognition,establishes the identification model of traffic jam mode of urban road network,and studies the forecasting method.The main work and innovations of the paper are:(1)This paper presents an identification framework and representation method for urban road network traffic.Based on the correlation of each factor in the road network and the importance of each road in the evolution of road network traffic patterns,the weighted K-means method is used to cluster the historical monitoring data of the road network,and the historical data are discretized and modeled by the results of cluster.This method is convenient for traffic managers to understand the real-time dynamic and changing rules of road network from micro and macro aspects,so as to effectively manage and make decisions and improve road network capacity.(2)This paper presents a forecasting method for the identification of traffic jam mode.Under the specific application background and identification framework of urban road network traffic jam mode,the pattern change and the time distribution rule of the road network are analyzed in depth,and three probability transfer matrices with predictive properties are obtained in this paper.By improving the Dempster combination rule in the theory of evidence,the information fusion method is used to fuse the probability transfer matrix,and the prediction result of traffic jam mode changein urban road network is obtained in this paper.The improved Dempster combination rule can reduce the complexity and redundancy effectively in the calculation process and improve the accuracy and real-time performance of the forecast.(3)This paper proposed a multi-mode congestion prediction method based on Pignistic interval length.Aiming at the shortcomings that the evidence theory can not effectively integrate the high conflict evidence,the Pignistic interval length is introduced as the measure of the distance between the evidence and the applicable scope of the evidence theory is discussed.Based on the idea of discount rate,the basic probability distribution function of each evidence is transformed to reduce the degree of conflict between the evidence.Finally,the improved Dempster combination rule is used to predict the traffic jam mode and improve the accuracy of the forecast model.