| Under the background of the dual carbon goals,the transportation industry is vigorously promoting the adjustment of the transportation structure and accelerating the realization of the "road to rail" of cargo transportation.However,in my country’s freight transportation market,the volume of road freight still accounts for more than70%.Excavating the potential of goods suitable for railway transportation(goods suitable for railway transportation)in road freight transportation,it can provide refined support for the " road to rail " strategy.With the rapid development of Internet technology and information technology,freight platforms and enterprises have accumulated a large amount of road and rail freight waybill data,which provides conditions for the mining of goods suitable for railway transportation.Based on this,this paper plans to build a training model based on large-scale railway waybill data and road waybill data to learn the characteristics of freight transportation in railway waybill data.The model is used as a suitable railway cargo discrimination model to intelligently identify and mine cargo suitable for railway transportation in the road waybill data.The main contents include:1.Research data preprocessing and basic analysis.This paper intends to study the classification of the types of goods by road and rail transport in the public and rail waybill data,as well as the distribution analysis of the basic characteristics of freight mileage,freight load and freight time in freight transportation.It is found that there is a problem of inconsistency in the classification of goods in the two transportation modes of public transportation and railway transportation.There are obvious differences in some freight characteristics of the two transportation modes,which can be used as the basic input characteristics of the cargo discrimination model suitable for railway transportation.2.Unified handling of road and railway cargo types.In order to analyze the characteristics of the types of goods in road and rail waybills,it is necessary to solve the problem that the types of goods transported by road and rail are not uniform.This paper studies the current classification standards of goods in the road and railway industries,and constructs a set of unified labels for road and railway goods based on the actual situation.It solves the problem of asymmetric classification of road and railway goods,and unify the characteristics of goods type input into subsequent models.3.Analysis on the characteristics of road and rail freight.In order to select the input features suitable for the railway transport cargo discrimination model,on the basis of the unification of cargo types,combined with the current research and the data of this paper,it is proposed to select the characteristics of freight mileage,freight weight,and freight time for in-depth integration and comparative analysis.Since whether the goods are suitable for railway transportation is not only affected by the characteristics of the freight,but also related to the regional railway transportation infrastructure capacity,this paper analyzes the differences in the regional railway transportation infrastructure capacity.The results show that the freight characteristics selected in this paper have an impact on the choice of transportation mode,which can be used as the subsequent input of the model.4.The construction and realization of the cargo discrimination model suitable for railway transportation.Using single-classification detection algorithms such as OCSVM,Iforest and CAE,the features excavated are used as model input to build the cargo discrimination model suitable for railway transportation.The three algorithms are compared and analyzed,and the convolutional autoencoder model with higher prediction accuracy is selected to predict whether a certain cargo category under multiple constraints is suitable for railway transportation.The model is used to identify the suitable iron goods source in the highway data,the results show that mineral building materials,cement,chemical raw materials and products have strong transit iron potential,while medical supplies,perishable goods,fresh animals and other goods have low transit iron potential.Guangdong,Sichuan,Henan and other places have a large amount of transit iron,while Xinjiang,Inner Mongolia and other places have very little transit iron.Moreover,it has a good positive effect on reducing carbon emissions and costs. |
