Analysis On The Propagation Mechanism And Prediction Method Of Urban Traffic Congestion Based On Multi-source GPS Data

In recent years,with the rapid economic and social development of China and the continuous improvement of people's living standards,the number of urban motor vehicles has continued to rise,and the contradiction between traffic demand and supply has become increasingly prominent.The resulting traffic congestion problem has also been seriously affected by the sustainable development of the city.To relieve traffic congestion effectively and enhance the standard of communications management,it is crucial to understand the formation process and propagation mechanism of traffic congestion.However,in the current situation,people lack sufficient and accurate understanding of propagation laws of urban road traffic congestion,and the management of traffic congestion mostly adopts afterward grooming strategy,which lacks the targeted and effective alleviation and prevention means.Therefore,this paper utilizes the data-driven analysis of the congested road section propagation mechanism as a point of penetration,proposing an urban traffic congestion propagation mechanism analysis and congestion prediction method based on multi-source GPS data.The research mainly includes the following aspects:(1)Collect and sort out massive historical GPS data of taxis and buses and smart card data.(2)Utilize Spatial Statistics Theory to assess the stability of road traffic congestion and find a recurrent traffic-congested road section.(3)Select a typical traffic congestion area,use long-period traffic spatiotemporal data to establish a traffic congestion propagation model,and calibrate the model parameters by calculating the traffic congestion propagation probability between related road segments,thereby constructing a reliable traffic congestion propagation model.(4)According to the analysis results based on traffic congestion propagation mechanism,combined with the similarity of real-time traffic status and historical traffic status and as well as traffic congestion graph model,to predict the possibility of traffic congestion for key road segments.Based on the above research content,this paper utilizes the taxi GPS,bus GPS,and smart card data from January to June 2019,and with the help of Hadoop and Map Reduce technology,builds a massive vehicle data processing platform and establishes a set of methods to find frequent traffic congestion sections from multi-source GPS data,constructing traffic congestion propagation model and calibrating.On this basis,traffic congestion prediction is carried out in the area of frequent traffic congestion,with the accuracy of the prediction results of more than 80%.The results of the research show that the traffic congestion propagation model constructed by the method proposed in this paper can truly and reasonably characterize the propagation features of traffic congestion both in time and space,and carry out effective traffic congestion prediction.Therefore,it has wide application prospects in the intelligent analysis and resolution of urban traffic congestion and the optimization of urban traffic management design.