Neural models predicting traction power consumption: support of e-energy telematic systems

Abstract

This paper deals with determination of relevant input data for models predicting electric traction power consumption. Furthermore, it deals with developing such models, which combine knowledge of classical physics with methods of artificial neural networks. It is possible to design control processes of railway traction effectively, based on these models. Real data of electric power consumption of traction substations in the Czech Republic within one year of train diagram 2006/2007 validity are used for verification of the models. These models can build support for e-energy telematic systems.