Research On Travel Characteristics Of Freight Vehicles Based On GPS Data

For a long time,road transportation has always occupied a dominant position in China’s cargo transportation industry.Under the background of rapid economic development,the demand of freight transport has increased tremendously in various regions,and the scale of road transportation is also expanding.At the same time,with the rapid development of big data,machine learning,artificial intelligence and other technologies,it provides scientific and technological support for in-depth mining of trajectory data information and further exploration of freight vehicle travel rules.Therefore,based on the GPS trajectory data of freight vehicles,this paper applies the trip chain theory to the study of freight traffic.On the basis of identifying freight business points and main base depots,the trip chain of freight vehicles is constructed and the related characteristics of the trip chain are analyzed,and then the freight vehicles are clustered based on the characteristics of the trip chain.It provides a theoretical basis for the modeling of trip chain behavior of freight vehicles,and provides a reference for the segmentation of freight market and for freight vehicle operators to provide customized transportation programs for different types of vehicles.The main contents of this paper are as follows:(1)The identification method of freight business point and main base depot of freight vehicles is proposed.By cleaning the original GPS trajectory data,the effective trajectory data is output,which provides a reliable data basis for the following work.On this basis,firstly,the effective stop points of the freight vehicles are extracted,and further eliminate the stop interference caused by long-term traffic congestion,rest in the service area,refueling,and so on,and then identify the vehicle business points.Then,based on the spatial distribution of business points,the improved DBSCAN algorithm is used to identify the base depots of vehicles,and the main base depot of each vehicle is extracted.(2)The freight trip chains construction and corresponding trip chain features recognition calculation method are proposed.Taking freight trip chains as the starting point of the research,based on the identification results of freight vehicle business points and main base depots,the trip chains are constructed and the characteristics of freight trip chains are identified and calculated.Combined with the calculation results,the overall distribution of all vehicle freight trip chains in time and space is analyzed.Then,taking a single freight vehicle as the research object,the average trip chain characteristics are analyzed.The results show that there is strong heterogeneity in the trip chain behavior patterns corresponding to different freight vehicles.(3)Vehicle clustering and vehicle travel characteristics analysis based on trip chain characteristics.Based on the characteristics of trip chain,the difference of travel patterns among freight vehicles is studied.On the basis of determining the feature clustering index,K-means algorithm is used to carry out feature clustering research on freight vehicles.In order to ensure the effectiveness of clustering effect,the feature data of trip chain was standardized,and the number of clustering was determined by elbow method and silhouette coefficient.Then,trucks are divided into three categories based on trip chain characteristics,and travel characteristics of each vehicle group are analyzed according to the characteristic results,which can provide theoretical support for freight behavior modeling and freight demand prediction.