Place of Neural Network Methods for Predicting Power Consumption in Railway Transport

The article shows the relevance of forecasting electricity consumption using an automated electric energy metering system that takes into account various levels of the hierarchy of the power supply system. The features of modern management of fuel and energy resources of the enterprise are considered, using the example of the Trans-Baikal Directorate for Energy Supply ‒ a structural subdivision of Transenergo ‒ a branch of the open joint stock company "Russian Railways". The rational use of fuel and energy resources, the introduction of energy-saving and resource–saving technologies are among the most relevant both in Russia as a whole and in railway transport in particular. The specifics of Russia's economic development and the specifics of the formation of tariffs in the electric energy market have led to an increase in its cost. In this regard, reducing the cost of purchasing fuel and energy resources is one of the main goals of the energy strategy of railway transport. The result of the energy-saving policy of railway transport, in almost all components of the consumption of fuel and energy resources, excluding train traction, is a reduction in the cost of buying electric energy in the situation of reforming electric energy by entering the wholesale electric energy market (RE). It makes possible to reduce the cost of the consumed energy significantly, so the cost of electricity received by the railway from the EPR will be greatly lower than the cost of electricity received from the guaranteeing supplier ‒ the regional power system. The paper analyzes the factors affecting on the amount of electric energy consumption for train traction. Neural network models with high approximating ability are considered, which allow processing statistical information and performing predictive estimates. It is shown that multilayer neural networks should be considered the most acceptable for predicting power consumption. The method of neural network forecasting of electric energy consumption is considered.

Abstract licence: CC BY

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.