The Research And Application Of Neural Networks And Four Stages Predict Method In Railway Transport Capacity

Fortune capacity forecasting work is in the very important position in the earlier stage work of traffic engineering construction. Can speak , and do getting well calculating analysis and the application of data to fortune capacity that being is in progress that the traffic is build and the design is worked starts a little , and is the foundation that project item construction scope and operation were economically appraised , is evaluation essential factor and the key of item risk . Calculating the flow of passengers in at a specified future date is a variable that the change is bigger, and is hard to pledge to the accuracy of forecasting.Neural networks calculates that the Buddhism the aspects of things make up doing the acquirement with four stages , and the construction is go out the new volume of traffic forecasting model in this paper. With the Jiao-Ji Railway is put forward speed and is transformed to the example, the passenger transport constituting is measured the forecasting model and have done the application study. Among them distributing with the traffic of average rate of increase law calculation flow of passengers capacity; The flow of passengers is brought out in model law calculation with gravity; Constructing the rate sharing responsibility for building model calculation according to transportation resistance shifts the flow of passengers; Think over time value when the flow of passengers is brought out in the calculation. Moreover the model building theory that the grey multi-level was appraised the law has still been inquired into in this paper, and has given out the application living example. This paper is appraised on theory foundation of law at the multi-level grey, to bring full play of all evaluation information in different grey degrees, and discusses the procedure of model creation. Then it induces all evaluators' distributive information into one priority-vector of different grey degree and allocates it with single value so a comprehensive value is obtained for the project under evaluation .The decision of sorting project to be evaluated then can be carried out.Thesis rich and developed now available traffic fortune measuring forecasting method and railway scheme from theory and puts into practice than the method which selected , and his major work and innovation point is following :1. This paper is measured national's total output value, per capita income and population etc factor and the traffic flows of passengers and has been build the direct contact, thus carries on the flow of passengers forecasting. And the forecasting method of tradition has very hard been building these factors to the direct contact with the volume of traffic. The living example certification indicates that the neuralnetworks model of this paper possesses the good convergence with the error back-propagation algorithm, and can pledge the satisfactory mapping precision, and gain the forecasting result of ideal, and provides the reliable basis for the policy decision. The neural networks predicted method broke through the limitations of traditional forecasting method.2. Being calculated with four stages that the Buddhism the aspects of things make up. Not only the forecasting method of combination offspring think over the influence of factors such as politics, economy and population etc, but also think over the influence of factors such as the time, price of a ticket and distance etc.3. The multi-level grey is appraised the law and is compared with same method, and the information think over is more overall, and appraising the result more accurately with objectively, practical nature is stronger, and more is suitable in carrying on objectively the just synthesis appraising, and is favor of pushing on the most excellent standardization selected and makes policy of railway scheme to the railway scheme than quality and the fixed quantity quota which picked on.