Traffic Conflict Modeling And Safety Evaluation At Expressway Toll Station

In recent years,China’s highway mileage has been steadily increasing,and the highway has become one of the main choices for residents to travel.As a key section of the highway,the toll station has an important impact on the safety and smooth flow of the highway.As vehicles still maintain high speed and frequent lane changes when entering the toll station diversion area,the toll station area is very likely to cause traffic conflicts,which implies great traffic safety hazards.Therefore,it is important to evaluate the safety of highway toll stations.This paper takes highway toll station as the research object,and based on the data of vehicle payment at the toll station,modeling analysis and traffic safety evaluation of regional traffic conflicts are conducted,and relevant suggestions for safety level improvement are given.The main research contents are as follows.(1)Toll station data collection and traffic characteristics analysis.Firstly,the collection of vehicle traffic data and regional environment data is completed by means of toll data extraction and on-site investigation.Then,considering the problem of inconsistency between the classification of vehicle models in toll data and the common classification,the vehicle models were reclassified according to the traffic survey classification standard.Considering the problem of realism in the simulation of conflict generation,the road environment characteristics and traffic operation characteristics of the toll station are analyzed separately and the parameters of the simulation are set in the later paper to ensure the reliability of the simulation conflict generation model.(2)Analysis of traffic conflict formation mechanism and conflict characteristics.Aiming at the substitutability of traffic conflict techniques in traffic safety research,the causes of conflict formation and the relationship with traffic accidents are first analyzed,and the effectiveness of using conflict techniques for alternative safety analysis is clarified.Then the conflict formation process is introduced,and the conflicts are classified into 2 and 3 categories respectively from both conflict perspective and severity.Then,considering the characteristics of the toll station area,the distance conflict time TTC is selected as the metric,and the conflict severity is classified into 3levels by the cumulative percentage frequency method.(3)Toll station traffic conflict model construction.Considering the problem of inconvenient conflict data collection,a conflict generation simulation model was firstly established based on Vissim,and the key parameters of the simulation were set based on the regional traffic characteristics.Then,for the problem of too many experimental scenarios that cannot be traversed,25 simulation scenarios are developed to obtain conflict data based on the L25.5.6 table in the orthogonal test method.Then for the total number of conflicts of all vehicles in the region,considering the discrete nature of conflicts,Poisson regression models were used to model and check the validity of the total number of conflicts,tailgating conflicts and lane change conflicts in the region respectively,and the highest R-square of the model reached 0.852,indicating a good fitting effect.For the conflict severity of individual vehicles,variables such as the vehicle’s own speed were selected,and various prediction models were established for the TTC values of vehicle conflicts based on machine learning methods,and the key parameters of each model were determined using genetic algorithms.Compared with other models,the prediction accuracy of decision tree is improved by up to 11.61%,which has better prediction effect.(4)Traffic safety evaluation model construction and safety level analysis for toll stations.Firstly,for the problem of weak comparability of the indicator of total conflict number,the conflict rate and serious conflict rate are selected as evaluation indicators.Then the safety level was divided into 4 levels based on the gray clustering model,and the toll station safety evaluation method was constructed.Then the model was used to evaluate each of the 25 simulation experimental scenarios,and the best scenario was determined under 5 traffic volume taking values.Then single-factor impact analysis was conducted for each of the 6 influencing factors,and suggestions for toll station safety improvement were made accordingly.