Detection and Mitigation of False Data Injection Attacks in Wide-Area Damping Controller Using Random Forest Classifier

Modern wide-area power system depends on communication-based operations for monitoring and control and are thus susceptible to cyber-attacks which can undermine the performance of WADC and their stability. Since the input signals for WADC are transmitted to the control location via communication channels, they are also vulnerable to cyber-attacks. An attacker could potentially inject false data to corrupt WADC measurements and control signals. Thus, this study focuses on developing attack models and examining the effects of false data injection (FDI) attacks on WADC performance. Further, a Random forest algorithm-based binary classifier is proposed to detect and classify the FDI-attacked signal from the non-attacked ones. The control framework for assessing such attacks incorporates local power system stabilizers (LPSSs) attached to each generator to mitigate local modes and a residue-based WADC along with the thyristor-controlled series capacitor (TCSC) tasked with damping critical modes within Kundur’s test system.