Application of Artificial Neuron Network in Analysis of Railway Delays

Punctuality is a key performance indicator of train freight transport. However, train delay arises often in the practice. To improve the efficiency of cargo train, prediction of train delay is always an important research area. In this paper, a prediction model is established on the base of artificial neural network (ANN). Due the endogen drawback of ANN, Genetic Algorithm is adopted to improve the performance of ANN. Consequently, an experiment is design to train and test the ANN-based model by a set of data from the practice. The results of the experiment demonstrate the significant propagation ability of the model.