Using Machine Learning Methods for Industrial Control Systems Intrusion Detection

In recent decades, information technology has been integrated into industrial control systems (ICS). At the same time, there was a connection of the ICS to the Internet and a transition to cloud computing. Consequently, new vulnerabilities and threats to sophisticated cyberattacks have emerged that create significant risks for the cybersecurity of ICS, and the old security model based on the isolation of ICS is no longer able to ensure their cybersecurity. This situation makes it very important to intellectualize the cybersecurity of ICS, for which machine learning (ML) methods are used. The use of ML methods will make it possible to detect cybersecurity problems of ICS at an early stage, as well as eliminate their consequences without real damage. This paper discusses the issues of ICS intrusion detection based on ML methods. The work can help in the choice of ML methods for solving anomaly detection problems of ICS.