Research On Private Vehicle Trajectory Generation Based On Spatial-temporal Interaction

Internet of vehicles is closely related to big data.The complex vehicular social networks can be constructed by the trajectory data,including social vehicles(such as private cars,etc.)and floating vehicles(such as taxis,buses,etc.).In contrast with the condition that the trajectory datasets of floating cars can be easily obtained from the Internet,it is almost impossible for general researchers to get the trajectory data of social vehicles because of personal privacy and some related government policies that are viewed as a hindrance for in-depth research and developments in related fields.Therefore,it is of far-reaching significance and value to study the practical method of generating data set of private car trajectory for smart city construction,traffic flow planning and prediction.Based on the theory of human travel spatial-temporal interaction,this paper aim to build two models of private car trajectory dataset generation.(1)From the perspective of the type and purpose of human travel,the vehicle trajectory dataset generation model based on urban regional representation is constructed.This paper proposes the Adjacent Road Segmentation method to divide cities into functional areas.At the same time,we move on to analyze the urban vehicles mobility patterns and propose the Regional Population-weighted Opportunities model.Then,the complete private vehicle trajectory datasets are generated by predicted urban travel volumes.(2)From the perspective of interaction behavior and travel decision game,we build the vehicle trajectory dataset generation model based on decision behavior.This paper that absorbs the thought of game theory considers the influence of individuals in the cluster while adopting the group game-oriented travel decision utility model to train from the existing taxi historical data to predict the inter-regional traffic in the future.Taking the impact of congestion effects into consideration,we adopt simulation tools and generative adversarial networks to train the trajectory prediction model so that the private car trajectory datasets conforming to social rules are generated.This paper is on the basis of the Beijing taxi trajectory data,point of interests data,and urban statistics.The research scope covers the area within the fifth ring road of Beijing.Finally,the model is constructed from macroscopic and microscopic perspectives to verify dataset generation methods proposed in this paper.The results show that the generated data not only has high accuracy and use value,but can provide strong data support for the Internet of Vehicles and transportation research work.