SVM Prediction Of Railway Freight Volume Based On Data Preprocessing

Railway freight volume is a key indicator to measure the competitiveness of railways.Accurate forecasting and grasping of freight volume and its development trend can provide an important basis for the development of railway transport organizations and transport management plans at all levels,and provide reliable support for the development of railway freight transport data support.At present,the research methods of railway freight volume forecasting at home and abroad are more in-depth,and the prediction performance and applicable conditions of various models are different.Therefore,reasonable prediction methods should be selected according to the objects to be predicted,and the advantages of each forecasting method should be fully utilized.Guarantee the reliability of the forecast results.SVM(Support Vector Machine)is a learning method based on statistical learning theory.It has good prediction stability and can solve the regression problem of nonlinear,small sample and high dimensional data.It can be used for railway traffic prediction.This paper does research work from two aspects: on the one hand,it first analyzes the general situation of China's railway freight transportation from a qualitative point of view,expounds the relevant quantitative indicators of railway freight volume,and analyzes the quantitative indicators,the freight categories,and foreign countries.Compared with other modes of freight transportation,the characteristics of railway freight transportation are summarized.The factors affecting the macroeconomics,logistics environment,railway itself and other factors affecting railway freight transportation and the correlation between various factors are summarized,and important influencing factors are selected.On the other hand,the SVM is introduced to quantitatively predict the railway freight volume.In order to improve the prediction accuracy of the traditional SVM prediction model and weaken the related factors of railway freight transportation,the data is preprocessed in two ways based on the traditional SVM model.Railway freight volume SVM prediction model is based on fuzzy information granulation method and phase space reconstruction method.By analyz-ing the freight situation of the Beijing Railway Bureau in recent years,the freight volume of the Shijiazhuang freight center and the coal traffic volume data of the Beijing freight center are collected to verify the model,and finally by comparing with other forecasting methods and traditional SVM methods,Conclusion: Compared with the unprocessed SVM prediction model and other methods,the accuracy of the two data pre-processed SVM prediction models is improved and can be used for railway freight volume forecasting.