Intellectual Diagnostics of the Blocks Electrical Centralization System at the Station
Abstract
In this article, the intellectual diagnostic system for the Block Electrical Centralization (BEC) system at railway stations was modeled using a performance-dependent artificial neural network (ANN) approach, fulfilling the requirements of the technical regulations of Uzbekistan Railways. The proposed system enables accurate and timely detection of various faults occurring in the station infrastructure without the need for direct human intervention, thus improving operational efficiency and safety. The ANN-based model was designed to monitor the six primary electrical circuits of each executive block within the BEC system. Voltmeters connected in parallel to each circuit provide real-time data, which is then analyzed by the neural network to identify anomalies and localize faults. The model architecture includes multiple hidden layers to enhance pattern recognition capabilities, even in complex operational scenarios involving simultaneous faults and high traffic loads. Experimental results demonstrated that the diagnostic system achieved a fault detection accuracy exceeding 96%, with a response time of less than one second. Moreover, the system proved to be robust against electrical noise and thermal fluctuations, ensuring reliable operation in real station environments. The graphical user interface developed alongside the diagnostic model allows electromechanical service personnel to visualize fault locations clearly and take corrective actions quickly. This significantly reduces maintenance times and minimizes the risk of service disruptions. Overall, the proposed ANN-based diagnostic system contributes to enhancing the safety, reliability, and economic performance of railway operations. It provides a scalable and practical solution for modernizing railway station infrastructure and supports the transition toward intelligent transport systems.