Comparative analysis of the effectiveness of neural networks and classical statistical methods in the diagnostics of microprocessor-based centralization systems

Abstract

The introduction of microprocessor-based centralization systems (MPC) is a key stage in the modernization of railway infrastructure, ensuring increased traffic safety and station throughput. Unlike morally and physically obsolete block route-relay centralizations, which are characterized by rigid logic, significant operating costs, and a lack of self-diagnostic capabilities, MPCs offer flexibility, high reliability, and integrated control, monitoring, archiving, and diagnostic functions.