Spatiotemporal Trajectory Reconstruction Algorithm Of Urban Expressway Vehicles Based On Data Fusion

With the accelerating process of urbanization and the continuous expansion of urban scale,the traffic demand increases sharply,and road transportation,traffic control and traffic safety are also facing great pressure.At the same time,with the diversified development of traffic detection technology,vehicles also produce a large amount of traffic flow data in the process of driving.Using data mining technology to analyze traffic data,explore deep-seated vehicle spatiotemporal trajectory characteristics,accurately grasp the road traffic operation state,and carry out real-time and accurate traffic management and control is of great significance to improve the performance of urban traffic system,improve the operation efficiency of urban traffic system,and maximize the interests of travelers and society.Taking the urban road network GIS data,floating car GPS data and AVI(Automatic Vehicle Identification)data as the research object,this paper proposes a vehicle spatiotemporal trajectory reconstruction framework based on data fusion,and puts forward a full spatiotemporal vehicle trajectory reconstruction algorithm by using the spatiotemporal correlation between the three data.Firstly,taking three typical traffic data,including urban road network GIS data,GPS data and AVI data as the research object,this paper introduces in detail from the aspects of data source,data characteristics and data preprocessing methods,and analyzes the correlation between the three data from the three dimensions of time,space and identity,which provides data support for vehicle spatiotemporal trajectory reconstruction.Secondly,a spatiotemporal trajectory reconstruction method of floating car based on interpolation algorithm is proposed.Based on the correlation between AVI data and GPS data in time,space and identity,the candidate path set of map-matching is simplified,and the correction method of GPS trajectory point coordinates for data fusion is proposed.The trajectory data is transformed from the actual geographical coordinates to the spatio-temporal coordinate system to obtain the discrete spatio-temporal trajectory of the floating car.The principles and characteristics of different interpolation algorithms are studied,and the missing or unknown trajectory segments of the known trajectory are repaired,so as to restore the complete spatiotemporal trajectory curve of the floating vehicle.Finally,according to the spatiotemporal trajectory curve of floating car,the estimation method of traffic flow parameters is studied to provide data support for the reconstruction of vehicle trajectory in the whole space-time.Then,a vehicle spatiotemporal trajectory reconstruction framework based on data fusion is proposed.The vehicle sequence variable is introduced into LWR model and Newell model,and the change of vehicle sequence is used to characterize the Overtaking Behavior of vehicles.The mathematical relationship between vehicle sequence and space,time and cumulative number of vehicles is deduced.The trajectory reconstruction problem is transformed into the solution problem of vehicle sequence,and a vehicle space-time trajectory reconstruction algorithm considering overtaking behavior is proposed.The algorithm uses the traffic flow parameters estimated by the reconstructed spatiotemporal trajectory of floating car as the input parameters,and the reconstructed trajectory of floating car is also used as the verification data set.Finally,a case study is conducted on some sections of Airport South Road in Shenzhen.The track reconstruction data set is established based on the GIS data of Shenzhen road network,the GPS data of taxi and the AVI data of video bayonet.The space-time track of taxi in some sections of Airport South Road during the evening peak period of September 1,2016(17:30-19:30)is reconstructed and visually displayed.According to the reconstructed taxi spatiotemporal trajectory,the traffic flow parameters are extracted as the input conditions of the trajectory reconstruction algorithm,and the full-time vehicle spatiotemporal trajectory is reconstructed and displayed visually.After error analysis,the average relative error re between the reconstructed floating car trajectory and the observed trajectory is 7.10%,which verifies the effectiveness of the trajectory reconstruction algorithm proposed in this paper.