Development of Cyberattacks Detection Framework of Power Systems Using Optimal Features in Smart Grids

Smart grids (SG) are a kind of sophisticated cyber-physical system (CPS) that allows power systems to be operated and controlled remotely by integrating smart devices' communication capabilities into the grid. However, this integration reveals many shortcomings in conventional SCADA (supervisory control and data acquisition) structures, which could jeopardise security objectives and present a significant cyberthreat to the SG. Therefore, in order to prevent grid sabotage, early detection of cyberattacks is essential. Conventional techniques for identifying cyberattacks that take use of smart grid network characteristics include machine learning (ML) techniques. An effective and efficient security control method is put forth in this work to identify cyberattacks on the SG. To increase detection efficiency, irrelevant characteristics are eliminated using the correlation-based feature selection (CFS) method. Using the chosen best attributes, an instance-based learning (IBL) system distinguishes between typical and abnormal occurrences. Depends on a limited set of features extracted from power system measurements of SCADA, experimental findings demonstrate that the suggested method achieves a high detection rate.