Research On Road Network Optimization Method Based On Vehicle Trajectory Data

The rapid development of Internet technology and the emergence of ride-hailing services have enabled the traffic managers can collect a tremendous amount of vehicle trajectory data.Vehicle trajectory data provide more abundant data sources for traffic control.Vehicle trajectory data are the information such as the location,time and speed of sampling points obtained by sampling the driving process of vehicles.Vehicle trajectory data have the advantages of wide range,real-time and low cost.Through the analysis and mining of trajectory data,this paper proposes a traffic signal control method for road network based on vehicle trajectory data.It divides the road network into control sub-areas and achieves coordinated control of trunk lines in the control sub-areas,so as to improve traffic efficiency of the road network.The main work includes:Firstly,this paper analyses the spatio-temporal characteristics of the trajectory data,and preprocesses the trajectory data to obtain an effective trajectory.By using the method of trajectory point superposition,the trajectory points of each phase of the intersection are accumulated to the same period,"spatio-temporal accumulative point graph"is designed through the velocity attributes of the trajectory point is used as the distinction.In the spatio-temporal accumulative point graph and time-distance diagram of intersections,the calculation methods of vehicle travel time,average parking time,average delay,maximum queuing length and maximum queuing dissipation time at intersections are put forward.The calculation method of queuing length at intersections is determined by analyzing the parking position and time of sampling trajectories,and the vehicle arrival rate prediction model based on poisson process is established by the arrival status and speed information of sampling trajectories at intersections.Secondly,this paper takes the ratio of vehicles passing through adjacent intersections in vehicle trajectory as the correlation coefficient of intersections,and establishes a control sub-area partition model based on the correlation degree of intersections.The corresponding traffic control parameters are calculated through the analysis and modeling of the trajectory data.With the optimization objectives of delay and parking time,and the saturation of intersection is constraint,the signal control cycle optimization model is established.Based on the characteristics of queuing dissipation and vehicle arrival rate,split optimization methods of pre-timed coordination control and actuated coordination control are proposed.Based on the analysis of queuing characteristics of traffic flow,an offset optimization model based on queuing length is established.Finally,built on the large-scale traj ectory data of Zhongguancun West in Beijing,through the traffic survey of 56 intersections in Zhongguancun West,the channelization status,the scheme of signal timing and motor vehicle flow of the road network in Zhongguancun West have been obtained.A real road network model based on Synchro is established,and taking capacity,delay,parking rate and queue length as evaluation indexes,the optimized scheme is compared with the existing scheme.The results demonstrate that the method reduces the average delay and queue length in the road network,and significantly improves the traffic of the road network.