NEURAL NETWORK PROCESSING
OF ASYNCHRONOUS MOTOR STATE PARAMETERS IN CASE OF INTER-TURN CLOSURE 

Dmitry V. Mirosh, Master of engineering, postgraduate student of the sub-department of locomotives, Belarusian State University of Transport (34 Kirova street, Gomel,
Republic of Belarus) dimamiroshheat@gmail.com

Background. The technical diagnostic tools used today in the repair and maintenance of asynchronous motors are an important aspect of the functioning of the most important devices in enterprises. Asynchronous drive is used in many areas of human activity, in industry as well as in everyday life. Materials and methods. The materials of this article present the technology of using convolutional neural networks for the diagnosis of inter-turn circuits in three-phase asynchronous motors with a short-circuited rotor, based on the use of a graphical representation of the relations of energy characteristics. Results. Data on vibration studies of traction electric motors of diesel locomotives, as well as direct measurements of the current of the studied asynchronous electric motors, were used as material for testing the capabilities and testing of the neural network. Conclusions. The use of the developed neural networks allows to improve diagnostic studies for asynchronous machines  of various capacities, easily adapt them to different dimensional designs, improve the quality of diagnostic services provided and reduce the labor costs of diagnostic specialists in the study of the parameters of the state of an electric machine.

 convolutional neural network, asynchronous electric motor, diagnostics, inter-turn closure, stator winding, artificial closure, traction rolling stock, electric motor connection
scheme, diagnostic efficiency, energy efficiency

Mirosh D.V. Neural network processing of asynchronous motor state parameters in case of inter-turn closure. Modeli, sistemy, seti v ekonomike, tekhnike, prirode i obshchestve = Models, systems, networks in economics, technology, nature and society. 2024;(3):105–115. (In Russ.). doi: 10.21685/2227-8486-2024-3-9