Research On Nonlinear Intelligent Forecasting Method For Passenger Flow And Revenue Of High-speed Railway Operation

At present,many countries in the world are building high-speed railway like fire and tea.The high-speed railway investment enterprises in some countries can repay the debts of construction and operation after many years of operation.During the period of operation of Japan’s Tokaido Shinkansen,the investors’ income has been in a good situation,and this line is also the most profitable high-speed railway.However,the passenger flow of high-speed railway built in most countries in the operation process has not reached the expectation,which makes most high-speed railway investment enterprises in the railway investment are in a state of loss.The number of passenger flow is not only one-sided measured from the population,high-speed rail fares,economic development level and other single factors,but is determined by a variety of complex factors.Based on the historical data of high-speed railway operation passenger flow and revenue,this paper studies the intelligent prediction model method suitable for high-speed railway operation passenger flow and revenue,so as to provide decision-making basis and support for improving the scientificity and accuracy of high-speed railway operation passenger flow and revenue prediction.In this paper,the original database of high-speed railway is established on the basis of collecting historical data of high-speed railway through various ways and methods,and the engineering characteristics that affect passenger flow and income of high-speed railway operation are analyzed and sorted out.Again using the SPSS data processing software to analysis of sorting out the engineering characteristics of relevance and independence test,based on the inspection results to build high-speed rail operating traffic and revenue forecasts based database information,then every high-speed lines operating history data traffic and income through regional adjustment coefficient,time adjustment coefficient to unify research base year and benchmark.Due to the influence factors on the high-speed rail traffic and the influence degree of different income,will affect the degree of small consideration will not only increase the cost of database also can make the calculation becomes more complicated,in order to solve this problem,this paper puts forward using reduction factors of rough set theory to construct high-speed operation database after the traffic and revenue reduction,BP neural network to need to have the certain learning samples before operation,in order to avoid the BP neural network algorithm in a local optimum in the training process,the learning samples for processing,namely from the underlying database by cosine similarity principle choose high iron engineering similarity with the target case,thus building suitable for the target database for the project,It provides reliable and scientific data support for subsequent BP neural network algorithm.The results of learning and testing the BP neural network model show that the prediction results of this model meet the prediction accuracy requirements of the feasibility study stage of high-speed railway project.Finally,Matlab is used to develop machine intelligent design and create the passenger flow and revenue forecast system of high-speed railway operation,which realizes a great progress in accurate,rapid and intelligent forecast of passenger flow and revenue of new high-speed railway operation.