Duration Time Analysis Of Non-Recurrent Traffic Congestion On Urban Expressway

In recent years,traffic congestion has occurred frequently,and the recurrent and non-recurrent traffic congestion has seriously affected people’s travel efficiency.Therefore,it is necessary to analyze the regular and estimate the duration of traffic congestion,while a driver chooses a travel path,or the traffic management department formulates a traffic control strategy.Based on the traffic information collected by the traffic information collection system for two weeks of Three Loop Route in Beijing,this paper uses the survival analysis method to study the distribution and the significant influencing factors of traffic congestion duration,compares the distribution of the duration of traffic congestion.The main content of the paper has the following points:(1)Analyze the overall distribution law of traffic congestion duration.On the basis of traffic information collection data,three different speed values are used as the threshold,from which three types of congestion samples are extracted,the recurrent and non-recurrent traffic congestion events are distinguished.Based on the non-parametric method of survival analysis,this paper estimates the distribution of the three samples including the overall congestion duration and the recurrent and non-recurrent traffic congestion duration.Wilcoxon method is used to test the distribution differences of overall traffic congestion and the duration of recurrent and non-recurrent traffic congestion duration under different spatial and temporal factors.The results shows that the non-recurrent traffic congestion lasts longer than the recurrent traffic congestion,and the probability of the end of the traffic congestion is lower after the congestion continues for a period of time.(2)The congestion duration distribution is analyzed based on the survival analysis parameter model.Based on the traffic congestion duration data,the accelerated failure time model is used to construct the distribution model of recurrent and non-recurrent traffic congestion duration.The lognormal model has the best fitting effect for the duration distribution of recurrent traffic congestion,while the ggamma model had the best fitting effect of non-recurrent traffic congestion after test the goodness-of-fit of the model.We can get the conclusion that the duration of non-recurrent traffic congestion is affected by spatial factors,while the duration of recurrent traffic congestion is affected by spatial,time and road conditions.In the case of non-recurrent traffic congestion,the inner loop lasts longer than the outer loop and the probability of the end of the traffic congestion is lower.Road environments such as bus stops,viaducts and ramp exits will increase the duration of frequent congestion,but have no significant impact on the duration of non-recurrent traffic congestion.(3)The risk factors of different traffic congestion states are analyzed based on the competitive risk model.Different congestion states in a congestion event are taken as the object of observation,and the transition from congestion state to non-congestion state is regarded as an event of interest,constructing a congestion duration competitive risk model and analyzing the duration and risk factors when the level is transitioning from congestion state to non-congestion state.The results show that the failure rate of mild congestion is the highest,followed by severe congestion,and finally moderate congestion,that is,after a period of congestion,the probability of mild congestion changing to non-congestion is the highest.