Queue Length Detection Of Road Intersection Based On Large Scale Vehicle Trajectory Data

The study of the queue length can form an effective measures to detect and prevent traffic congestion.It also has a positive effect on alleviating the pressure of road network and promoting the healthy development of urban traffic.Vehicle GPS trajectory data(floating car data for short)which is generated when the vehicle run on the road is a kind of typical mobile data.It is a hotspot to obtain the length of the queue by analyzing and mining these floating car data.However,existing researches based on floating car data often has the problem such as strict application condition,parameter setting only for specific road intersections and the related research on the size of the data sample is lacking.With floating car data and network from Shenzhen,we put forward a method based on the theory of spatial distribution pattern to detect the queue length of vehicles at road intersections and identify the optimal sample size for queuing length detection to deal with existing problems.In this paper,the betweenness centrality and the vehicle flow of the road network are first calculated to obtain the road set with different properties.Next,this method analyzes the distribution of sampling intervals in the original floating car data and determine the uniform sampling interval of the trajectories for the queue length detection and the optimal sample size identification.According to the formation mechanism of the actual queue at road intersection,the track point characteristics of the queuing vehicles are extracted.Considering the fact that the floating car data is actually the track point left by the vehicle running on the road,the spatial distribution pattern theory is introduced to quantitatively obtain the spatial point pattern of track points.In this way the queuing length of the vehicle can be detected using the extracted characteristics and the spatial point pattern of the track point.Finally,the influence of the floating car data sample size on the detection results and the efficiency of the algorithm is analyzed,we use the kernel density estimation principle to obtain the density distribution curve of the track point under different sample size on the road,and measure the stability of the curve quantitatively,so as to confirm the optimal sample size of the intersection length estimation.The experimental results show that the queue detection algorithm proposed in this paper have strong applicability under the premise of ensuring high precision and can be used for the detection of queue length of multiple types of road intersections.In addition,the optimal sample size determined in this paper is highly correct,and the conclusion is concluded that the optimal sample size of the road intersection which are constituted by the similarity attribute road is consistent.With floating car data taken as the object of study,a queuing length detection algorithm with strong adaptability is proposed and the corresponding optimal sample size is determined in this paper.It provides a new method for the vehicle queue length detection of large scale urban road intersection.