СИСТЕМА ДІАГНОСТУВАННЯ ТЕХНІЧНОГО СТАНУ ТЯГОВИХ ЕЛЕКТРОДВИГУНІВ ТЕПЛОВОЗІВ ЧМЕ3 З ВИКОРИСТАННЯМ ШТУЧНИХ НЕЙРОННИХ МЕРЕЖ

Abstract

The tendency for practical implementation of automated diagnostic systems is meant to formalize, simplify, and automate the process of diagnosis. The device of artificial neural networks allows to solve tasks of estimation of efficiency both for an object as a whole and for its specific functionally separated systems. The results of the work of artificial neural networks allow to determine the degree and nature of the influence of operational factors on the development of the processes of damaging the elements of the traction electric machine, as well as to determine the impact of all stages of the operation of the locomotive (work under load, run, waiting for work, etc.) for the state of its isolation, bearings, collector node and other elements. This, in turn, allows you to determine in advance the degree of possible change in the characteristics of the isolation of the traction electric vehicle used and to make a conclusion on the need for preventive or repair work. The equipment of traction electric motors with such a system will help to prevent or at least minimize the failure of traction electric machines during operation due to the lower dielectric strength of isolation of its windings, disruption in the work of bearings and brush-collector node. It is offered to apply self-teaching neural networks in the diagnostics of the operation of traction electric motors of diesel locomotives CHME3 with the prospect of spreading the accumulated experience to all systems as a whole.