Publication

Abstract : In the cyber-physical era, machine learning plays a vital role in detecting and classifying cyberattacks to safeguard the systems. In this paper, the impact of False Data Injection (FDI) attacks is investigated on two-area Load Frequency Control (LFC) systems, with a focus on stability and operational resilience. This research study utilizes large datasets of 18,000 and 50,000 entries that reflect various operational states under normal conditions and potential FDI attacks, which are labeled as scaling, ramp, and random attacks. This research study strengthens the LFC system to withstand cyberattacks and enhances the overall power system resilience. A few of the Machine Learning (ML) classifier models like Support Vector Machines (SVM), the ExtraTrees with AdaBoost (ET+AdB), and the Principal Component Analysis (PCA) with (ET+AdB) are employed to train and test the dataset. These ML models enable the prediction and classification of anomalies in the proposed two-area LFC systems, facilitating real-time response to enhance the security and robustness of LFC operations. The performance of the ML model on the system is evaluated with metrics of accuracy, precision, recall, and F1-score. Among the three classifiers, the (ET+AdB) classifier demonstrates superior robustness in detecting and classifying attacks. The generated diverse datasets of 18k and 50k can be accessed from the IEEE data port.