Analysis the Informative Value of Digital Images of Signals from Industrial Control Systems Under the Influence of Cyberattacks

The focus of this article is on the information content of a digital image of an observed time signal from an industrial control system (ICS) that is affected by cyberattacks. The process of creating a digital image of the observed signal involves three stages: obtaining a time-frequency image of the signal, performing two-dimensional digital filtering of the time-frequency image, and selecting the most informative components of the image. An analysis of process behaviour and detection of anomalies in ICS processes, which may occur during cyberattacks, is conducted using machine learning technologies and deep neural networks applied to the digital image of the observed ICS signal. The research identifies anomalies in observed ICS processes by analyzing both the signals and time-frequency images of the ICS signals. The article shows that detecting anomalies from time-frequency images of ICS signals produces better results than detecting anomalies directly from the ICS signals. This text highlights the informative nature of digital time-frequency images of ICS signals in detecting anomalies caused by cyberattacks.