Predictive maintenance of integrated security and communications system for passenger trains

The article discusses the possibility of applying the concept of predictive maintenance, namely the possibility of creating models to prevent failures of integrated security and communications system used by passenger trains. Various methods for constructing models for predicting equipment failure are considered, namely the technical equipment operation simulation, neural network technologies, machine learning. Building a model for predicting hardware failures using simulation and neural network technology is problematic. It is not possible to collect statistical data because of the wide geographical spread of the equipment. The most promising direction is machine learning, which includes several stages: data collection, analysis of accumulated data, creating models to predict failures, preventing failures. Building a model to predict hardware failures will help to determine a viability stage of equipment. It is necessary for the shutdown maintenance of the equipment or for equipment pre-ordering. The economic impact of the proposed model is defined by reduction in storage and maintenance cost, due to the fact that repair cost is smaller than new equipment cost.