Font Size: a A A

The Research Of Railway Passenger Flow Forecasting Method Based On Greyneural Network Models

Posted on:2017-03-13Degree:MasterType:Thesis
Country:ChinaCandidate:M XianFull Text:PDF
GTID:2272330485975235Subject:Traffic Information Engineering & Control
Abstract/Summary:PDF Full Text Request
In the evaluation system of railway transportation,railway passenger flow is an important index, railway passenger flow’s data and trend forecasting can be used to the adjustment of train working diagrams and directlyaffect the reliability and practicability of the railway transportation dispatching system,so how to accurately predict the railway passenger flow is particularly important. Due to objective factors,the railway passenger flow is oscillatory sequence.According to this feature,prediction models are constructed based on the grey system theory and the BP neural network to adapt to the nonlinear characteristics of railway passenger flow and to enhance the practicability of forecasting model and the forecasting precision.Based on the ability of information processing on small sample data,the two modelshave a high utilizationin many fields.This paper will have an in-depth study on both them and propose improvement program tobuild prediction models used in the railway passenger flow prediction.And Eventually get the optimal prediction model in the form of combination forecast.Through the analysis of grey technology, the traditional grey prediction model is not suitable for nonlinear data prediction,so this paper will use the buffer operator and the gray power model to predict.The buffer operator can restore the original visage of distortion data,which is suitable for all kinds of data sequences including nonlinear data.The gray power model can get good prediction accuracy for nonlinear sequences,because the power index of the gray power can adjust forecasting curve according to the characters of the original data. Constructing the two models with real data proves the two forecasting models are better than the GM(1,1) model.The BP neural network have a strong learning ability,but there maybe occur some negative effectsin the traditional BP neural network training,example stagnant learnning,local minimum and so on.According these shortcomings,learn the current improved algorithm and put forward an optimizationprogram based on adaptive learning vector in this article to reach the purpose of convergencing network quickly and efficiently,finally complete the model by MATLAB.The combination with the two models can remedy the deficiency of single forecasting model and get good ability of data processing and forecasting.Study combination and optimization method and build the prediction model to prove the combination method’s goodforecasting and stable practicability.
Keywords/Search Tags:grey system, traffic flow, nonlinear sequence, BP neural network, combination forecast
PDF Full Text Request
Related items