Risk Assessment Of Expressway Weaving Sections Based On Trajectory Data

With the rapid development of China’s urbanization process,the number of vehicles in urban areas continues to increase,the level of urban expressway construction continues to improve,and traffic safety issues have gradually come to the fore.Especially in the intertwined area of expressways,vehicles travel fast and often change lanes frequently,and the traffic flow is complicated,resulting in traffic accidents from time to time,and the safety problem in the intertwined area is becoming increasingly serious.Considering the difficulty of obtaining accident data,we use traffic conflict data instead of accident data for statistical analysis and conduct a study on the risk assessment of the intertwined area.Firstly,manual survey and UAV aerial photography are conducted in the interweaving area of the expressway to determine the characteristic parameters of the geometric structure of the interweaving area,obtain the video of traffic flow running in the interweaving area,select the YOLOv5+Deepsort algorithm to detect and track the vehicle target on the aerial photography video,and further combine the coordinate conversion method to obtain the vehicle position,speed,and other trajectory information data.Secondly,we analyze the mechanism of traffic conflict in the interweaving area of the expressway,select ETTC as the traffic conflict index according to the reason for conflict,count the traffic conflict in the study area with 5s as the time window,obtain the conflict matrix,the number of conflicts and the id of the conflicting vehicles in a single time window,and use the 85%cumulative frequency method to classify the interweaving The traffic conflict severity level in the area is divided into three levels.The number of conflicts,the severity of conflicts,and the number of conflicting vehicles are used as independent variables,and the K-means algorithm is used to cluster to obtain the traffic risk status of a single time window within the survey period in the intertwined area.based on the study of the operational characteristics of traffic flow in the interweaving area,27 variables related to the risk of the interweaving area are identified from three regions:interweaving area,mainline,and ramp,and the Pearson correlation coefficient method is used to test the variables for multicollinearity,and SVC-RFE and RF-RFE are selected to rank the variables in terms of importance and construct the evaluation of traffic risk in the interweaving area The set of indicators was constructed.The risk state(no risk=0,risk=1)of the intertwined area is used as the explanatory variable,and a binary logit model is used to construct a risk identification prediction model for the intertwined area.The maximum likelihood method was used to estimate the number of covariates,and four and eight models were fitted based on the forward likelihood test and the backward likelihood test,respectively,and the Cox&SnellR2 and NagelkerkeR2 were used to compare the model fitting.The results showed that the seventh model fitted by the backward likelihood test was the best,with a prediction accuracy of 76.0%.Finally,OR values are introduced to qualitatively and quantitatively analyze the correlation between each evaluation index and the risk status of the interweaving area,and combine the correlation to analyze the reasons and mechanisms of each evaluation index affecting the risk status of the interweaving area,and propose risk prevention measures as well as management suggestions from three aspects,such as speed,congestion status,and others.The risk prediction model of the interweaving zone can diagnose the risk of the interweaving zone of the urban expressway,evaluate the safety of the interweaving zone as a whole,and provide risk warning for the vehicles that will enter the interweaving zone to ensure the safety and smooth traffic flow of the interweaving zone as a whole.In addition,the study also proposes the process of calculating and extracting vehicle trajectory data based on the aerial video of UAV,which has certain practical guidance significance.