Prediction Of Traffic Volume At Intersections Based On Trajectory Data Of GPS Floating Car

With the rapid development of Global Position System(GPS)and communication technology,floating car,which is also called probe vehicle,has changed the traditional way of traffic information collection.Floating car can provide real-time traffic information such as vehicle's location and speed.In addition,GPS floating car trajectory data can provide us with more comprehensive road traffic information.It can provide continuous spatial information of vehicles.From the GPS trajectory data,we can know what happened during the driving process of vehicles.Recently,using GPS floating car trajectory data to estimate traffic state and behavior has been developed rapidly.However,most of the existing studies mainly focus on using GPS trajectory data for traffic parameters estimation such as average speed and travel time of road or queuing length and delay at intersections,and then estimate whether the traffic condition is smooth or congested,without quantifying the traffic volume specifically.At signalized intersection,traffic volumes are the fundamental and key inputs to many signal optimization algorithms.It is significant to estimate traffic volume in intersection using GPS floating car trajectory data for optimizing signal timing.In this paper,we mainly established the intersection trajectory prediction model based on the speed-density relationship,and then predict the traffic volumes in intersection by GPS floating car trajectory data.Firstly,the paper introduced the theoretical basis of the model,including the concept of traffic flow parameters and several commonly used speed-density models.In this paper,we proposed the model based on the speed-density relationship,to estimate the vehicles between the floating cars,and then we can obtain the traffic volume.According to the definition of speed and density,in the study range of space-time,the traffic speed can be approximated by the average of all floating car speeds in this period,and the traffic density can be approximated by the mean of instantaneous density at different moments during the observed period.Secondly,the paper established the intersection trajectory prediction model based on speed-density relationship.Due to the influence of signal lights at intersection,the driving process of vehicle is complicated,and divided the road sections according to the different driving states of vehicles.Then the paper established the prediction models based on the trajectory data before the vehicle enters the intersection and the trajectory data when the vehicle is released,which are used to predict the number of vehicles between the floating cars in the same cycle and the vehicles between the queued floating cars in the same cycle.Next,the paper introduced the traffic volume prediction method at intersection based on GPS floating car trajectory data.Since the proposed intersection trajectory prediction model is used to predict the vehicles between floating cars in the same period,it is necessary to judge whether two adjacent floating cars are in the same period before the prediction.Then the paper predicted the vehicles between floating cars in the same period by the intersection trajectory model.And then we predicted the vehicles between the stop line and the first queued floating car before the stop line.Based on the predicted values obtained above,the traffic volume of intersection is estimated.Finally,the paper established the intersection simulation model based on SUMO and generate the vehicle trajectory data.The model is tested using SUMO simulation data.The paper used the error of traffic volume as the evaluation index to estimate trajectory data with penetration rates of 20%to 50%.