Application of artificial neural networks for electric load forecasting on railway transport

The article is devoted to the use of artificial neural networks for electric load forecasting of railway transport. It has considered approaches to load forecasting for electric rolling stock and stationary objects of railway transport. It has performed the analysis of the main factors influencing the consumption of electricity for rail transport, and elaborated the mathematical model of power consumption using artificial neural networks. Proposed additional criterion for assessing the quality of the neural network model based on the F-Fisher test. On the basis of the proposed algorithms has developed a software package for predicting the consumption of electrical energy and made his approbation on railway transport of Russia.