Defects recognition of axle caps of the rolling stock wheel-motor block based on the results of modeling an artificial neural network for predicting output diagnostic parameters

The article presents the results of the research conducted by the authors, the purpose of which was to development a model for recognizing defects in axle caps of a wheel-motor block of a locomotive in order to implement automatic advance notification of management structures about the need for maintenance or repair operations to eliminate defects at an early stage of their occurrence. The research used the following interdisciplinary and mathematical methods: computer and mathematical modeling, methods of mathematical statistics, methods of the theory of artificial intelligence and parametric reliability. As a result of the research, a mathematical formalization of the model for recognizing one of the defects in the axle caps of the wheel-motor block of the locomotive - the groove (chipping) of the babbitt layer was obtained. With the help of the obtained model, it is possible to implement automatic recognition of defects, pre-failure states not only of axle caps, but also of other units of technical systems. The developed model can be used in monitoring systems, control, diagnostics of the technical condition of the locomotive fleet, in order to reduce downtime in repairs and forced costs for scheduled operations. The proposed model solves the range of problems described in the development concept of JSCo Russian Railways associated with the implementation of the actual repair system for the current technical condition of the locomotive, as well as with the digitalization of the company's advanced areas.