Research On Characteristics Of Logistics Vehicle Delivery Behavior Based On Trajectory Data

With the continuous expansion of the development scale of the logistics industry,the number of logistics delivery vehicles on urban roads continues to increase,and the delivery behavior of vehicles presents more and more complex characteristics.Research on the distribution behavior characteristics of logistics distribution vehicles based on trajectory data can help logistics companies monitor vehicle distribution behaviors,efficiently allocate various logistics resources,solve the problems of scattered urban logistics facilities layout and unclear functions,and improve logistics distribution efficiency.Based on the trajectory data of logistics delivery vehicles,this paper uses descriptive statistical methods,kernel density analysis,spatial autocorrelation analysis and clustering algorithms to analyze the delivery behavior of vehicles from the three perspectives of distribution time and space characteristics,distribution space patterns and distribution area characteristics.The main research work is as follows:(1)From the perspective of logistics and distribution vehicles,excavate the temporal and spatial characteristics of vehicles.With the aid of box plots and cumulative frequency graphs combined with scale-free characteristics,the temporal and spatial statistical laws of distribution vehicles are preliminarily analyzed from the three perspectives of distribution time,distribution distance and distribution range.Using the method of line graph and kernel density analysis,combined with the effective period of delivery vehicles,the time and space distribution characteristics of logistics delivery vehicles are analyzed.The results show that 10:00-12:00 and 14:00-16:00 are the peak hours of distribution travel,and the distribution vehicles are concentrated on the Jingtong Expressway and the Chaoyang District Logistics Distribution Center.A number of distribution vehicles load,unload,and enter and leave Beijing.(2)From the perspective of urban travel areas,excavate the distribution pattern of vehicles.Using Thiessen polygons to divide the urban road network,61126 urban travel areas are obtained.Based on the delivery distance of the vehicle in different travel areas,using three exploratory spatial data analysis methods of SA,HLC and LISA,combined with Moran’s I index,Getis and Ord G index,p-value and z-score,derives the distribution space model of the vehicles.The results show that the distribution behavior in the urban travel area roughly presents a spatial pattern of high concentration at the periphery of the Fifth Ring Road,low concentration in the central city and local travel areas outside the Fifth Ring Road.(3)From the perspective of the logistics distribution area,excavate the distribution behavior characteristics of vehicles in each area.Based on the time series of vehicle delivery distances in each travel zone,three measurement methods: Euclidean distance,DTW distance,and improved DTW distance based on the attenuation coefficient are used to calculate the distance between time series to obtain the corresponding time series similarity matrix.When the direction of change between two time series is opposite,the time series similarity matrix based on the above distance can’t effectively reflect the true similarity between time series.For this reason,the CORT method,which is improved based on the adaptive adjustment function,is used to modify the above-mentioned matrix.Based on the modified matrix,using TS-PAM clustering algorithm,six types of logistics distribution areas are initially obtained.According to the distribution distance characteristics and POI distribution characteristics of different distribution areas,the distribution areas are further integrated into four types of distribution areas.Finally,the distribution characteristics of the four types of distribution areas are summarized.