Mechanism Analysis Of Urban Traffic Congestion Based On Fusion Of Multi-Source Data

With the development of economy and society in China,the traffic industry has also been developed rapidly.The demand and quality of residents’ travel are constantly improving.The current traffic supply cannot meet the traffic demand of hot spots in peak period.The resulting traffic congestion not only seriously affects the daily travel quality of residents,increases the travel cost,but also induces traffic accidents and causes environment Pollution and increase of resource waste are not conducive to the sustainable development of China’s economy and society.The existing research puts forward many different governance schemes and measures from the perspective of traffic supply and demand to solve the urban traffic congestion problem.However,to explore the deep causes of traffic congestion,we should also implement the relationship between traffic and land use,and solve the problem of mismatch between traffic supply and traffic demand from the root,so as to solve the traffic congestion problem.In this paper,POI is fused The paper analyzes the spatial and temporal distribution characteristics of traffic congestion,studies the relationship between traffic and land use from macro and micro perspectives,forecasts the congestion state of different land use combinations by BP neural network.Finally,according to the research on the relationship between traffic and land use and the prediction results of neural network,the corresponding treatment is put forward The measures are reasonable.Firstly,the POI data and traffic status data of electronic map are collected in the open interface of electronic map by Python program.The relevant data can be obtained in real time,accurately,efficiently and low-cost.On this basis,the rules cleaning,classification and merging of data are processed to improve the accuracy and availability of data fusion and data analysis,and provide the data basis for the research.Secondly,on the basis of preprocessing the POI and traffic status data,the paper analyzes the characteristics of traffic congestion time distribution through the calculation of evaluation indexes on the macro level;through the spatial matching of POI and traffic status data,the land use nature around the congestion point is analyzed;through the comparison of land use around the congestion points at different time points,the paper analyzes the congestion prone to congestion at different time points Land use.In micro view,K-means clustering is used to determine the cluster center by using the distribution characteristics of land use properties reflected by different types of POI data.The improved network kernel density model can reflect the characteristics of "distance attenuation effect" that influence intensity gradually decays along the shortest path in spatial distribution,and represents the radiation effect relationship between cluster center and congestion point within the threshold range;the cluster center can be used to determine the cluster center Based on the model,the coupling model of traffic and land use is constructed to reflect the influence of different land use types on the congestion point,identify the land use type which has the greatest impact,and classify the congestion points based on this.Finally,based on the relationship between traffic and land use from macro and micro perspectives,the congestion state of different land use combinations is predicted by BP neural network,with the prediction accuracy of 84.821%.According to the analysis of land use distribution combination around different traffic states,the paper studies the influence of different land use layout combinations on congestion.Through neural network prediction and different land use layout combination,the corresponding congestion control measures are put forward.This paper makes full use of and explores the big data resources of network traffic,and comprehensively and accurately analyzes traffic congestion,which provides the basis for traffic congestion management.