Building an expert system for diagnosing traction electric motors of rolling stock

Abstract

This article is devoted to the development of an expert system for diagnosing traction electric motors of rolling stock. The article describes the process of selecting diagnostic parameters, allowing to assess the quality of machining and assembly of the traction motor. The process of formation of a fuzzy model of diagnosing a traction motor based on the Takagi-Sugeno algorithm is shown. Selected diagnostic parameters are taken as input linguistic variables of a fuzzy model. As an output linguistic variable, a complex indicator of the quality of mechanical processing of the collector of a traction motor was adopted. The article presents the membership functions for the input and output linguistic variables and the rules for fuzzy products for the model being implemented. The hypersurface of the output linguistic variable in the space of various attributes is shown. Based on a fuzzy model, a neural fuzzy network has been developed, which allows to evaluate the quality of mechanical processing of a collector of a traction motor, and presents the results of its training. The developed neural fuzzy network is recommended to be used as one of the components of a comprehensive expert system for diagnosing traction electric motors of rolling stock.