Machine learning based diagnostic for SCADA systems

Energy distribution systems and Cyber-Physical Systems have a very important place in terms of information technology. The use of information technology in these areas also poses a risk. SCADA systems used in these systems will become cloud-based systems that can communicate with IoT devices in the future. But SCADA carries quite a risk of attack over time. Therefore, both the detection and prevention of these attacks are of great importance. In this thesis, attack threats are detected and classified using machine learning algorithms against SCADA systems in Cyber-Physical Systems. While making this classification, a unique value is assigned to each attack type. In this study, as well as attack detection, the type of attack is determined with this unique value assignment. Detection of the attack, determination of its type and performance results were carried out in the test environment and the results were obtained.