Optimization and integration method for railway freight stations based on a hybrid neural network model

Yan Sun, Maoxiang Lang, Danzhu Wang

School of Traffic and Transportation, Beijing Jiaotong University, Haidian District, 100044 Beijing, P.R. China

Given to the current problems existing in the operation of railway freight stations and the entire railway freight transport network, in order to integrate the railway freight stations and optimize the traditional railway freight transport mode, we first propose a strategy on the optimization and integration for railway freight stations, then design a hybrid neural network model to recognize the operating performance of each railway freight station by classifying them into four ranks based on the proposed strategy. The characteristic of the proposed model is its combination of the respective advantages of unsupervised learning algorithm based neural network and supervised learning algorithm based neural network. Finally, an empirical study from Hohhot Railway Administration is given to verify the feasibility of the proposed model. The simulation results of the empirical study indicate that (1) the accurate recognition of training samples has significant influence on the classification result; (2) the proposed model can recognize the operating performance of the railway freight stations under relatively high accuracy.