Model-based fault detection and isolation for wind turbine

In this paper, a quantitative model based method is proposed for early fault detection and diagnosis of wind turbines. The method is based on designing an observer using a model of the system. The observer innovation signal is monitored to detect faults. For application to the wind turbines, a first principles nonlinear model with pitch angle and torque controllers is developed for simulation and then a simplified state space version of the model is derived for design. The fault detection system is designed and optimized to be most sensitive to system faults and least sensitive to system disturbances and noises. A multi-objective optimization method is then employed to solve this dual problem. Simulation results are presented to demonstrate the performance of the proposed method.