کنفرانس مهندسی برق ایران

صفحه اصلی / سی و چهارمین کنفرانس بین المللی مهندسی برق

AI-Driven Quantitative Fault Diagnosis and Adaptive MPC for Wind Turbine Icing and Actuator Faults

نویسندگان :

Reza Bidari¹ Moosa Ayati² Yassin Riyazi³ Farzad A. Shirazi⁴

1- University of Tehran 2- University of Tehran 3- University of Tehran 4- University of Tehran

کلمات کلیدی :

Wind turbine،Fault-tolerant MPC،Blade icing & actuator faults،Quantitative fault diagnosis،Adaptive Neuro-Fuzzy Inference System (ANFIS)،Empirical Mode Decomposition (EMD)

چکیده :

In this paper, an integrated fault diagnosis and control framework is proposed for a variable-speed wind turbine operating under scenarios involving blade icing or actuator faults. The proposed scheme combines data-driven fault detection with a fault-tolerant Model Predictive Control (FT-MPC) strategy to maintain optimal performance and system stability, providing a foundational framework capable of identifying distinct fault types. First, an adaptive signal decomposition and statistical feature extraction approach is used to generate fault-sensitive features. These features are then fed into an AI-based estimator to quantify fault severity in real-time. Then, the quantitative fault information (e.g., icing severity and actuator effectiveness) is incorporated into the MPC controller through an adaptive model update mechanism that adjusts control constraints and prediction horizons accordingly. The control objectives include maximizing aerodynamic efficiency, minimizing power fluctuations, and ensuring structural safety under faulty conditions. Simulation studies conducted on a 5-MW NREL wind turbine benchmark demonstrate that the proposed scheme effectively detects faults in real time and maintains stable operation with reduced pitch angle and generator torque deviations compared to conventional MPC and fault-tolerant control methods.