Resilient Lyapunov-Based Model Predictive Control for Wind Power System Under False Data Injection Attacks

Wind power systems operating in networked environments are vulnerable to stochastic disturbances, measurement noise, model mismatch and false data injection (FDI) attacks. These uncertainties may corrupt feedback information and degrade closed-loop control performance. This paper proposes an integrated extended Kalman filter (EKF)-based resilient Lyapunov model predictive control (RLMPC) framework for the secure control of wind power systems under bounded stochastic FDI attacks. A residual-based chi-square (χ2) detector is embedded into the EKF update to evaluate the credibility of received measurements, and the resulting attack-aware state estimate is applied to the RLMPC controller at each sampling instant, constructing an EKF-RLMPC strategy. The proposed EKF-RLMPC scheme therefore links attack detection, state estimation, and predictive control within a unified secure-control framework for wind power systems. It is proved that the posterior estimation error remains bounded and that the closed-loop state is ultimately bounded under the proposed EKF-RLMPC scheme. Simulation studies under different FDI attack probabilities show that the proposed method improves state-estimation accuracy and control performance.