Study On The Division Of Comprehensive Economic Regions And The Characteristics Of Highway Freight Transportation Based On SOM Algorithm

As one of the important ways of transportation in our country,road transportation is the main way of carrying goods in our country.Because highway transportation has its own characteristics such as time autonomy,flexibility and so on.Moreover,the density of highway network is higher than that of railway network and waterway network,which can provide "door-to-door" short distance transportation,which is the bridge and tie of economic connection between regions,and has the guiding function to economic development.Similarly,road transportation is developed on the basis of regional economy,but the economic characteristics of different regions are different,and the demand for road freight transportation will be different in scale,structure and level.In order to make reasonable regional road transportation planning,feasible regional transport strategy and give play to the role of guiding economic development,we should use the economic and facilities related to road freight to carry out regional division research.In order to provide a reference for the development of regional road transportation,we seek areas where economic development and infrastructure construction are similar to each other,so as to formulate specific freight transport policies according to the freight transport characteristics of different regions.Based on the statistical data and the time series data of active level of highway freight transport,this paper selects the economic,facility and highway accessibility indexes related to road freight transport,and constructs a spatial clustering model.After the division of the region,the characteristics of regional road freight transport are analyzed.Based on Matlab programming software and R language,the author uses Matlab to write self-organization mapping net(SOM)algorithm model and R language to measure the similarity of freight transport trend.Firstly,we refered to the selected indicators in the existing literature,taking the data availability,science and accuracy as the principle,and normalizing them as the data basis of the next research.Secondly,the selected indexes are screened and eliminated.Through spatial autocorrelation and multi-multiple collinearity analysis,the indexes with agglomeration characteristics and no correlation between data are obtained,and then the geo-weighted regression model is constructed.Analyze the interpretation ability of the selected indexes to the road freight related data.Then,the cluster model is applied to cluster analysis,and combined with the principle of regional division,it is preliminarily divided into nine categories,but the results do not conform to the principle of division of provinces.After further clustering analysis of industrial structure,the provinces with special conditions are classified.In order to verify the rationality of the classification of special provinces,the dynamic time warping(DTW)algorithm is used to measure the similarity between the special region and its neighboring provinces,which further proves the rationality of the classification of special provinces,and get the seven economic regions Finally,based on the economic regional division and analysis of regional road freight transport characteristics,the distribution of freight transport and the size of factors affecting regional road freight are analyzed.It can provide some reference basis for different regions to formulate the future road freight development planning goals and so on.