Mining Traffic Congestion Propagation Patterns Based On Spatio-temporal Co-location Patterns

With the acceleration of urbanization and the popularization of motor vehicles,city traffic congestion has become a common social problem.There is a widespread phenomenon of congestion,that is,congestion occurs in a certain road section and spreads to the surrounding areas.Because of the propagation characteristics of traffic congestion,it is easy to cause largescale traffic congestion,resulting in safety problems,and even make the city suffer from air pollution and noise pollution.Urban traffic congestion has become a key problem in urban development,and it is also one of the main problems that municipal departments need to solve urgently.Due to the complexity of traffic roads and the propagation of congestion,the congested road section will have a direct or indirect impact on the road network in the traffic congestion event.Evacuation of traffic congestion for local sections alone cannot achieve good results.And the impact of different congested roads on the overall network is different.Therefore,in a traffic congestion event,different treatment schemes for traffic congestion will produce different congestion propagation effects,identify the propagation paths that play a key role in congestion propagation,and effectively manage the relevant road sections is an effective way to alleviate urban traffic congestion.Based on this idea,this paper proposes a congestion propagation pattern mining algorithm based on spatio-temporal co-locations.Based on the traditional co-location pattern mining algorithm,the temporal and spatial characteristics of traffic data are considered.This paper aims to mine a kind of congestion propagation pattern which not only has enough prevalence,but also causes congestion to spread rapidly in the road network.First,we measure the spatiotemporal proximity relationship between congested roads based on their topological path distance on the road network and the time of congestion.Second,we divide the congestion data into congestion event sets according to the spatio-temporal proximity relationship,and measures the propagation influence of the congestion road on the congestion events with the spatio-temporal neighborhood affected by the congested road.Furthermore,we measure the interestingness of a pattern by its participation index and propagation influence,so that we can mine congestion propagation patterns that frequently co-occur and have a strong propagation influence.The proposed algorithm was experimentally verified on the Guiyang traffic dataset,and the experimental results revealed the traffic congestion law of Guiyang.Compared with the APSTCP algorithm,on the basis of mining prevalent co-occurrence traffic congestion patterns,it can also mine congestion propagation patterns with strong propagation influence.The accuracy of the algorithm is tested experimentally,and the top-10 mining results are selected for analysis,which verifies the effectiveness and practicability of the algorithm.