Data-diven Study On Accelerated Lane Length In Expressway Interchange Interchange Merging Zone

With the gradual advancement of the construction of urban expressways in China,the emergence of bottlenecks in the interchange area of expressways has also attracted increasing attention.The expressway interchange confluence area is prone to traffic bottlenecks,which not only seriously inhibits the traffic capacity of the interchange,but also affects the full play of the function of the expressway,and also causes a large number of traffic safety hazards.According to the measured data,the vehicle influx area is studied from the aspects of macro traffic flow and microscopic vehicle behavior,and the vehicle follow-up lane change model of different confluence zone sink behavior is trained by generating adversarial network,and finally the acceleration lane index design of the confluence zone is designed,and the recommended accelerated lane length design value is obtained through traffic simulation,the main results include:Firstly,the basic characteristics of the accelerated lane in the urban expressway interchange and merge area are expounded,and the vehicle traffic data of the interchange interchange and confluence area of some urban expressways in Wuhan are collected based on field research;according to the measured data,starting from the macro and micro levels,through the analysis of traffic flow characteristics,traffic conflict characteristics,vehicle convergence behavior,etc.,the difference between the vehicle follow-up lane change behavior in the convergence area and the general road section is obtained: ramp vehicles accept insertion gaps into the main line according to a certain probability.Mainline vehicles move lanes to the inner lanes in advance or slow down to avoid ramp vehicles.Then,according to the characteristics of traffic flow in the confluence area,the data agglomeration class is analyzed to classify different driving behaviors from the data level,and the vehicle follow-up lane change model is constructed based on the generated adversarial network,and the effectiveness and advantages of the adversarial network are analyzed by using different methods to train the vehicle follow-up lane change model.Finally,taking the length of each part of the acceleration lane in the merging area as the optimization object,the SUMO traffic simulation system obtains the traffic flow status of the merging area under different acceleration lane lengths,and evaluates the traffic flow by four indicators: speed,traffic density,rear-end collision rate and lane change conflict rate,and obtains the recommended acceleration lane length combination under the three-level service level: the recommended length of the accelerated lane under single-lane direct at 100km/h,80km/h,and 60km/h is223 m,207m and 194 m The recommended length of the single-lane parallel acceleration lane at 100km/h,80km/h,and 60km/h is 350 m,331m,303m;the recommended length of the two-lane direct acceleration lane at 100km/h,80km/h,60km/h is 752 m,694m,644m;the recommended length of the two-lane parallel acceleration lane is 1124 m,1055m,991 m at 100km/h,80km/h,and 60km/h.And compared with the acceleration lane index value in the "Urban Road Intersection Planning Code".The reasonable design of the acceleration lane length of the interchange and confluence area plays a key role in the traffic efficiency of the entire urban expressway system and the improvement of driving safety,and this paper adopts the method of traffic simulation based on data-driven and established vehicle follow-up lane change model to study the acceleration lane length design of the urban expressway interchange interchange convergence area.It has reference value for alleviating the frequent traffic congestion in the merging area and improving the driving safety level of vehicles in the confluence area.