Identifying Disturbances in the Behavior of Technological Processes in Intelligent Monitoring Systems

The paper examines a new method of creating intelligent monitoring systems for railway transportation management processes based on two classes of basic monitoring models designed to solve the main monitoring problems of assessing conditions and diagnosing disturbances in the behavior of the supervised process. The first class is a statistical model for analyzing multidimensional data flows based on the method of principal components that is designed for transforming the initial parameters of the supervised process into a number of generalized state variables for the purpose of their subsequent use as part of monitoring algorithms. The second class is a model of an evolutionary fuzzy system designed for extracting knowledge regarding the behavior of the supervised process from a data flow for the purpose of assessing its states and diagnosing disturbances. The basic monitoring models were used as the foundation of the concept and new method of intelligent monitoring based on two chains of data flow transformations. The key advantage of the developed method of intelligent monitoring is the in-depth diagnostics of classes of disturbances that enables the detection of disturbances, identification of their nature, direction and scale, as well as their causes.