Research On Trajectory Reconstruction Method Of Highway Based On Travel Time Estimation

The construction of smart highways can realize the rapid and intelligent development of the transportation industry.It is necessary to accurately estimate the running state of the road network,analyze the space-time characteristics of the vehicle in the process of building a smart highway.Vehicle trajectory data has individual space-time characteristic,which is helpful for the construction of smart highways,therefore,the research of this subject needs to be carried out.Taking the travel time as the transition parameter and using the travel time as the constraint,a vehicle trajectory reconstruction technology based on travel time estimation is studied.The main work and innovations of the thesis are as follows:Basing on the complex traffic modeling problem in traditional travel time estimation,the long training time and unstable effect in the travel time estimation based on neural network,the travel time estimation method based on feature matching is studied.The feature is built into the feature library,the feature matching is completed by the KNN search algorithm,and the travel time is estimated by the mean method.Aiming at the problem that the direct reconstruction in the traditional trajectory reconstruction method leads to large error,based on the travel time estimation,a trajectory reconstruction method based on the trust region algorithm is studied.The estimated speed of travel time is used as the initial input of the algorithm,and the real travel time is taken as the constraint.The nonlinear optimization model with minimum compensation is established to realize the trajectory reconstruction.Analyzing the influence of the cross-section detector spacing and the running state of the road network on the reconstruction accuracy,using the discrete Frechet distance between the reconstructed trajectory and the real trajectory as the basis for the similarity evaluation.The accuracy of the section detector data,bayonet data and reconstructed trajectory data for the estimation of the running state of the road network is analyzed and compared.The simulation results show that the similarity increases with the decrease of the geomagnetic sensor distance of the detector,and decreases with the congestion of the running state of the road network.The reconstructed trajectory can estimate the road more accurately than the data of geomagnetic sensor and traffic bayonet.The research results in this paper can effectively support the construction and management of smart highway traffic,and provide a new direction for the application of highway traffic trajectory in big data environment.