Deep Machine Learning Model-Based Cyber-Attacks Detection in Smart Power Systems
Energy supply and consumption can be effectively managed while avoiding a number of security risks thanks to modern intelligent energy grids. Events that are both man-made and naturally occurring can disrupt a system. Operators must understand the various types and reasons behind disruptions in the energy systems in order to respond and make decisions that are well-informed. In order to address this issue, this study suggests a deep learning-based attack detection model for energy systems. This model can be trained using logs and data collected by phasor measurement units (PMUs). Features are created using property or specification creation, and data is sent to different machine learning techniques, with random forest being chosen as AdaBoost’s default classifier. The model comprising 37 energy system event case studies is tested using publicly available simulated energy system data. Ultimately, the recommended model has been contrasted with alternative designs using a range of evaluation criteria. The results of the simulation demonstrated that this model outperforms the current methods, achieving a detection rate of 93.6% and an accuracy rate of 93.91%.
