A Communication Failure and Cyber-Attack Resilient Wide-Area Out-of-Step Prediction Scheme Using Dynamic State Estimation

Out-of-step (OOS) prediction is essential for reducing mechanical and electrical stress on power system components. A reliable strategy involves using wide-area measurement systems (WAMS) to monitor the dynamic behavior of voltage angle differences across transmission lines and generator rotor angles. However, the limited availability of phasor measurement units (PMUs) presents a challenge, as voltage angle measurements are not available at all buses. Moreover, rotor angles cannot be directly measured and must be estimated. This paper proposes an OOS prediction scheme based on dynamic state estimation (DSE), which concurrently estimates both algebraic and dynamic states. By calculating the slip frequency and the acceleration of voltage angle differences, the electrical center's location is predicted by tracking its trajectory on the slip-acceleration plane. A similar approach is used to predict generator OOS based on the estimated rotor angles. Since the scheme depends on WAMS data, it is susceptible to cyber-threats such as false data injection attacks (FDIAs), denial-of-service attacks (DoSAs), and communication failures. To improve resilience, the proposed method integrates a penalized weighted least squares (PWLS) approach into an extended Kalman filter (EKF) for detecting and mitigating attacks and handling data loss. The scheme is validated on the IEEE 39-bus test system under various network topologies and attack scenarios.