Development of Cybersecurity Systems Based on Machine Learning and Deep Learning Algorithms for Industrial Control Systems

Industrial control systems, often known as ICSs, are frequently utilized in serious infrastructure because of their widespread use in the support of key societal processes. This is because ICSs are able to monitor and regulate a wide variety of processes. As a result, the protection of these systems from cyberattacks is very necessary. Over the course of the last several years, a variety of distinct ideas for numerous kinds of cyberattack detection systems have been offered, with every concept making use of its own unique assortment of processes and methods. A substantial gap exists in the research on techniques for detecting cyberattacks in ICSs, and this research gap has to be addressed in order for the study to be considered complete. The key purpose of the proposed system is to recognize instances of cyberattacks that target ICSs. We describe an artificial intelligence-based approach for abnormality identification in industrial control systems (ICSs). The purpose of the technique is to act as a guide for any future research that may be conducted in this field. The suggested algorithms were evaluated on actual ICS datasets. The suggested system was developed in two phases, the first of which used binary classification and the second of which involved multiclass classification. The binary classification was used to determine if the malware was an attack or a typical instance, whereas the multiclass classification was responsible for identifying each unique attack. In both binary and multiclass classification, the DT and KNN algorithms obtained an accurateness of one hundred percent, which is the highest possible. As per the findings of the sensitivity study, both the DT and KNN algorithms attained R2 values of 100% throughout both phases. The classification provides 95.7% accuracy.