The imperfection of systems for monitoring and analysing the state of hard-to-reach devices and nodes, dynamically flowing technological processes associated with them necessitates the analysis of not only “internal” data, but also those arriving through external and external channels. This article discusses one of the most effective ways to solve this problem by using digitized data sequences grouped into behavioural patterns. The analysed device, as a rule, is outside the controlled area, and access to it is limited, as a result, all processes occurring inside the devices must be under constant control. Changes in temperature, amplitude and frequency of vibrations, sound, and electromagnetic spectra taken by sensors and sensors during the production and operation of individual components may indicate the need for changes in the technological process in order to prevent wear, breakdowns, and rejects. Thus, the analysis of the state of hard-to-reach nodes and devices of Industry 4.0 based on incoming data from external side channels is an urgent task. The essence of the proposed approach is to obtain the dependence of the quantitative indicators of the functioning of devices, infrastructure nodes of "Industry 4.0" for various modes of operation and further use of the obtained data when solving the problem of analysing the state of devices and infrastructure nodes of "Industry 4.0". The novelty of this study is the development of an approach to the analysis of Industry 4.0 devices designed to identify the state of remote and hard-to-reach devices, which combines machine-learning technologies in the analysis of heterogeneous external data of digitized trac es of signal sequences coming from various elements and combined into behavioural patterns.
Read full abstract