Intelligent condition monitoring of railway signaling in train detection subsystems

Abstract

Condition monitoring is a prevalent method to improve the Reliability, Accessibility, Maintainability and Safety level of a system. By using this method, detection and diagnostics of imminence or persistence faults, will be viable. The ability of detecting imminent faults, it will be possible to prevent any breakdowns via preventive maintenance, which leads to improve Reliability and Accessibility and functional lifetime expansion of the system. In recent years, monitoring systems have gained much consideration for boosting the quality of equipment operations in railway systems. In this paper, we employed Neuro-Fuzzy Network NFN for fault detection and diagnostics in a typical audio frequency track circuit, which is a combination of knowledge-based and data-based systems. It has the merits of both fuzzy systems and neural networks. In other words, NFN is capable of interacting with low precision data and has the learning ability of a neural network. For the realization of the method; a typical track circuit has been modeled and simulated. Healthy and faulty data have been used for training the algorithm. When put to work, occurrence of fault modes or their imminence is detected and localized with a good precision by the algorithm.