A method of using neural fuzzy models to determine the technical state of a diesel locomotive’s electrical equipment

Abstract

Equipment for carrying out onboard diagnostics is an important component in systems for controlling the state and reliability of rolling stock, in particular, that of diesel locomotives. It makes it possible to monitor a locomotive’s service quality, to detect and predict changes in its technical state without interrupting the carriage process, and to efficiently use the high-cost stationary and local diagnostic means, as well as to correct service time and volume by taking into account the real technical state. Therefore, it is important to develop methods for determining the technical state of locomotive units by using an MCS-T(N,E) (micro-processor control system) locomotive’s diagnostic subsystem of a locomotive and software for processing the measurement data. The main aim of such a method is to monitor the technical state of a locomotive’s equipment and predict changes in it. At the present time, simplified diagnostic models are used to solve such problems. However, to increase the efficiency and reliability of the process of checking complicated objects such as a locomotive, it is necessary to use more complex diagnostic models based on artificial neural networks. A way to use a neural-network method and neural fuzzy diagnostic models for diagnosing the excitation system of traction generators of modern diesel locomotives is presented in this paper. Information collected by the MCS-TP subsystem is used as diagnostic information. It is transferred online to a remote diagnostic server. The method for diagnosing the excitation system of a traction generator is checked by processing the measurement information.