Study On The Forecast Of High-speed Railway Passenger Traffic Volume Based On Combined Model

In recent years,in order to support China's economic development and respond to the rapidly expanding transportation demand,China has begun to enter a period of large-scale investment and construction of passenger dedicated lines,and the construction of large-scale railways,especially high-speed railways,has eased the current tense passenger transport capacity,which will bring great convenience to railway passenger transport.The accurate prediction of high-speed railway passenger volume plays an important role in the economic evaluation of high-speed railway construction projects,the allocation of national resources,the adjustment of railway internal investment structure,and the management of operations.In terms of railway transportation,forecasting traffic volume is an important task that needs to be completed before the investment in railway planning.Therefore,scientific and reasonable prediction of high-speed railway passenger volume has become an important research topic in the transportation field.Because the high-speed rail passenger volume is affected by many factors,the data often exhibit a certain degree of volatility,and it is often difficult to accurately describe the objective laws.Different forecast methods have different effects on the accuracy of the forecast results.The information obtained by using a single forecasting method is limited.The related research shows that the combined forecasting method is obtained by combining different single forecasting methods in a certain way,which can improve the overall prediction of the model.The purpose of this paper is to introduce the IOWA combined forecasting method and the PSO-BP method into the high-speed rail passenger volume forecasting field.With high accuracy as the goal,the two types of combined models were studied separately for the prediction of high-speed rail passenger traffic.In order to improve the prediction accuracy of high-speed railway passenger volume,on the basis of expounding the characteristics of passenger traffic changes,this article analyzes a number of factors affecting high-speed railway passenger traffic,and finally determines the factors by using the correlation coefficient method.Secondly,taking into account the development changes of high-speed railway passenger traffic,we selected partial least-squares regression,GM(1,1)and BP neural network as a single predictive model to analyze and predict high-speed rail passenger volume,and then all single prediction models are combined by using IOWA operator to establish the IOWA combined forecasting model.Then the BP neural network and the particle swarm optimization algorithm are analyzed and summarized,and the defects of the BP neural network are pointed out.The PSO-BP model combining the particle swarm optimization algorithm and the BP neural network algorithm is proposed,and the model is simulated and predicted by using MATLAB.Finally,based on the historical data of China's high-speed railway passenger traffic and various influencing factors,the IOWA combined forecasting model and a combined model of PSO and BP network were established respectively.These two types of combined models were used to predict the high-speed railway passenger traffic in China,and the predicted value of high-speed railway passenger volume were got,finally four evaluation indicators are selected to judge the prediction performance of each model;the results show that the first three error indicator values of the two combined forecasting models are lower than the three single forecasting model,and the R value is closer to 1.Compared with the other three the single prediction method,the error of the combined model is smaller,and the prediction accuracy is higher,which verifies the validity and practicality of the two combined forecasting models proposed in this paper,thus providing an effective reference for traffic planning.