Research On The Method Of Freight Train Distribution Situation Deduction Based On XGBoost Algorithm

In recent years,with the rapid development of China’s railway,the line mileage and transportation capacity are significantly increased,and the railway network is also being perfect.With the increase of railway network density and the construction of passenger dedicated lines,the separation of passenger and freight has been realized in some areas,and the freight transportation capacity has been greatly improved.Meanwhile,a well-organized real-time railway transportation production information dynamic collection and reporting system is established,which is capable of producing a large number of railway freight train relevant data.Whereas,these freight train data are not further processed and analyzed,and there is no grasp of the whole railway network transportation production.Under this background,this thesis aims to study the situation of freight train distribution based on the relevant data of railway freight transportation,and with the method of the XGBoost algorithm,optimized by the max-plus algebra method.The study is beneficial to both theoretical research and practical application for the macro control of the evolution laws of the whole railway network transportation production.The main contents are shown as follows:(1)The main factors,which have influence on the railway freight train distribution,are discussed from various aspects.Combined with the existing railway freight statistical index system,the new freight train situation index is constructed and the long-term trend and short-term trend of the situation index are analyzed,based on the characteristics of the railway freight train distribution situation.Two rounds of expert consultation are conducted according to Delphi method,and three different periods of railway freight train distribution situation index system are constructed.According to the 3-hour railway freight train distribution situation index system,the calculation method and situation level interval of each index are analyzed.(2)With the help of the max-plus algebra,the running process of freight train on the railway network is analyzed,and the static freight train running process model is constructed.According to the influence of incident on freight train operation deviation diagram,the dynamic buffer matrix is designed,the dynamic freight train operation process model is constructed and the solution algorithm is designed.(3)The advantages and disadvantages of various integration algorithms are discussed,the XGBoost algorithm is used to build the situation model of railway freight train distribution,and the accuracy and calculation speed of the XGBoost algorithm is compared with other integration algorithms and the max-plus algebra for the delay time deduction of freight trains.The XGBoost algorithm is improved by using the max-plus algebra to construct the abnormal data under the occasional event,which enhances the strength and toughness of the model.(4)Taking the real railway network situation of the Wuhan Railway Bureau under heavy snowfall as an example,the model is used to deduce the situation of railway freight train distribution of the Wuhan Railway Bureau,and a comparative analysis of the results is carried out to verify the effectiveness and practicability of the model.