Modern Approaches to Diagnostics and Recovery of Rolling Stock Information Management Systems Using Multi-Agent and Neural Network Technologies

The article considers modern approaches to diagnostics and recovery of the rolling stock information management systems (IMS). Introduction: in the context of increasing demands on the transport system reliability and safety, traditional diagnostic methods are becoming ineffective. Purpose: to explore the possibilities of using multi-agent and neural network technologies to improve the IMS diagnostics and recovery efficiency on the rolling stock. Methods: the paper analyzes multi-agent systems that provide distributed diagnostics where agents interact for malfunction detection and localization. Neural network technologies that ensure high accuracy of fault prediction through the analysis of large amounts of data and self-learning are also being investigated. Results: the main result is to identify the potential for integrating multi-agent and neural network technologies. Their combined use can significantly increase the reliability, adaptability and autonomy of the rolling stock IMS. Discussion: the prospects of creating hybrid systems combining multi-agent and neural network methods are discussed. Their advantages and limitations are considered and the potential for further improvement of transport system IMS is illustrated.