The roundabout intersection has traffic self-organization ability,which can reduce the severity of vehicle conflicts and improve traffic safety.With the increase of motor vehicle ownership and the rapid increase of urban traffic flow,the shortcomings of non-signal-controlled roundabouts are gradually revealed.At present,China adopts the optimization method of non-signal-controlled roundabouts for reconstruction or signal control management,so as to solve the problem of congestion and bottlenecks at roundabouts.In this case,the actual measurement study and accurate simulation model of the actual roundabout intersection is an important reference value for the optimization of non-signal-controlled roundabout intersection.And with the development of computer technology,the agent technology is gradually applied in the field of transportation,and the agent system can simulate the operation of the agent group in reality more accurately,and the application of agent system in urban traffic modeling and simulation has made some progress.Therefore,this thesis intends to carry out microscopic traffic simulation modeling of non-signal-controlled roundabout intersections with the introduction of agent technology,and the main research work and conclusions are as follows.Non-signal-controlled roundabout intersection data acquisition.Video of vehicle operation at the intersection is captured using an unmanned aerial vehicle,including the whole process of vehicle entering,circling and exiting the loop.Using trajectory extraction software and video processing software to extract the vehicle information data from the recorded video,the traffic volume,vehicle speed and headway time distance data of the roundabout intersection of Benma Road in Nanjing,Jiangsu Province were analyzed by actual measurement,and it was found that its roundabout traffic headway time distance could be fitted with a shift negative exponential distribution.The arrival of vehicles in the inlet lane obeys the negative binomial distribution;the design of the roundabout intersection plays the effect of reducing the speed of vehicles.Gap acceptance behavior data at the entrance of the roundabout intersection are extracted from the video.Based on the accepted gap values and rejected gap values,the critical gap is solved using the Raff method and the maximum likelihood estimation(MLE),and the results show that the critical gap at the Benma Road roundabout is lower than the recommended value of the critical gap given by the U.S.Highway Capacity Manual.Simulation model construction of non-signal-controlled roundabout intersection based on agent.Define the traffic in the simulation model scene module,vehicle agent module,information storage module,describe the role of different modules in the simulation and the functions of six modules inside the vehicle agent,analyze the dynamic process of the vehicle agent,and classify the vehicle agent according to its motion process,and build the generation model,the following model,and the lane change model.Focus on building the lane change model of the vehicle agent,classifying the vehicle lane change behavior into three categories: in-loop,around-loop and out-loop lane change,and modeling the motivation of each category of lane change behavior.Model calibration and validation.The parameters in the motion model are calibrated based on the measured data.The objective function is constructed,and the genetic algorithm in Python is used to calibrate the car-following model and the lane change model.The calibrated parameters are applied to the simulation model,and the performance indexes of vehicle speed,vehicle distance and displacement are used to verify the control effect of the following model and the lane change model on the vehicle.The headway time distance of the circular traffic is used as an index to verify the effectiveness of the overall model.Meanwhile,the accuracy of the VISSIM model and the model in the thesis is compared using the section speed and average travel time as indicators.The relative error of the model in the thesis is more balanced and the overall error is smaller than that of the VISSIM model,so the model constructed in the thesis is more applicable to the roundabout intersection of Benma Road. |