Prediction Of Railway Passenger Traffic Based On Time Series

With the rapid development of China’s economy,China’s railway industry has also been rapid development.At present,the main problem on China Railway is to make the railway transport capacity in order to satisfy the needs of national economic and social development.In the early stage of construction,it is necessary to accurately forecast the railway traffic in the planning year and provide a reasonable basis for the railway construction scale.Because the accuracy of the prediction depends on the choice of the prediction method.So accurate prediction on the future passenger traffic information has become the key to enhance the railway transport capacity.Firstly,the paper introduces the data and current development of China Railway in recent years,and explains the importance of the short-term traffic forecast and its significance to railway development.And then,it introduces the main forecasting methods and analysis their features by conventional forecasting methods and new forecasting methods.Then the paper also introduces the latest research in the use of these methods in the industry.And finally the truth is that most of the current traffic forecasting methods have some limitations.So it is necessary to find a scientific and simple way to predict short-term traffic.Through the analysis and research on the characteristics of railway passenger flow,the paper clarifies the value and significance of the railway weekly passenger flow forecasting work.It is found that the time series of railway weekly passenger traffic is a periodic and seasonal nonstationary time series.This paper analyzes the advantages and disadvantages of EMD model,ARIMA model and SVR model.Three EEMD-related combinatorial models are proposed by using the idea of combinatorial model——ARIMA model based on EEMD decomposition,SVR model based on EEMD decomposition and ARIMA-SVR model based on EEMD decomposition.The corresponding modeling steps and their characteristics are described respectively.Finally,the paper uses the train traffic data from Qingdao to Beijing train 2013 to 2015 as the research object。And then the short-term passenger flow forecasting model is established by EEMD-ARIMA,EEMD-SVR and EEMD-ARIMA-SVR respectively,and the results are obtained.Finally,it is concluded that the combination model has higher prediction accuracy than the single model in terms of the fitting of historical data and the prediction of the model;Secondly,the EEMD method has a good effect on the decomposition of time series.At the same time,it shows that EEMD-ARIMA-SVR has good adaptability and significance for passenger flow forecasting model.At the end of this paper,the research results are summarized and the direction of further improvement and further research is put forward.