Robust fault estimation of a blade pitch and drivetrain system in wind turbine model

Abstract

In this article, a novel robust fault estimation scheme to ensure efficient and reliable operation of wind turbines has been presented. Wind turbines are complex systems with large flexible structures that work under very turbulent and unpredictable environmental conditions for a variable electrical grid. The proposed observer-based estimation scheme consists of a set of possible faults affecting the dynamics, sensors, and actuators of wind turbines. First, the pitch and drivetrain system faults occur simultaneously with process and sensor disturbances that are called unknown input signals. Second, through a series of coordinate transformations, the faulty subsystem is decoupled from the rest of the system. The first subsystem is affected by unknown inputs, and the second one is affected by faults. A reduced-order unknown input observer is designed to reconstruct states accurately, whereas a reduced-order sliding mode observer is designed for the second subsystem such that it is robust against unknown inputs and faults. Moreover, the reduced-order unknown input observer guarantees the asymptotic stability of the error dynamics using the Lyapunov theory method and completely removes unknown inputs; on the other hand, the reduced-order sliding mode observer is designed to reconstruct faults for the faulty subsystem accurately. Until now, authors only focused on an unknown input signal in the dynamics of the system, especially in nonlinear systems. The estimated fault will be adequate to accommodate the control loop, and sufficient conditions are developed to guarantee the stability of the state estimation error. In the next step, to figure the effectiveness of the proposed approach, a wind turbine benchmark system model is considered with faults and unknown inputs scenarios. The simulation results are used to validate the robustness of the proposed algorithms under noise conditions, and the results show that the algorithm could classify faults robustly.