Article Title

Application of neural networks at simulation ofa system of current collection on electric railways

Journal: Journal of Transsib Railway Studies V.1(33), 2018
Journal thematic sections: Transport power engineering
Pages: 69-79
Authors: A. S. Golubkov
udk: 004.032.26
Download PDF
Article reference
Golubkov A. S. , Ermachkov G. R. , Smerdin A. N. , Sidorov O. A. Application of neural networks at simulation ofa system of current collection on electric railways Izvestiia Transsiba – The Trans-Siberian Bulletin, 2018, no. 1(33), pp. 69 – 79.

Abstract

The article deals with the analysis of methods for obtaining statistically reliable data on contact pressure, which are based on the results of inspection trips and data on the design values of the location of the contact wires in the vertical plane andin the path plan. A method for calculating contact pressing with artificial neural networks is proposed. The methods of obtaining statistically reliable data on the contact pressing of current collectors of electric rolling stock without the need for direct measurement, based on the video image of the current collection process and analysis of external factors (weather, operational) accompanying the interaction are considered.

Article keywords

Current collector
Neural network
Contact force
Mean-square error
Closed-cycle prediction

Authors

A. S. Golubkov —
Candidate of Technical Sciences, associate professor of the department «Electricity supply of railway transport», OSTU.
G. R. Ermachkov —
Post-graduate student of the department «Electricity supply of railway transport», OSTU.
A. N. Smerdin —
Candidate of Technical Sciences, associate professor of the department «Electricity supply of railway transport», OSTU.
O. A. Sidorov —
Doctor of Technical Sciences, Professor, Head of the department «Electricity supply of railway transport», OSTU.