Active fault tolerant control of wind turbines using identified nonlinear filters

This paper presents a scheme for accommodating faults in the blade pitch, rotor speed and generator torque control sensors of a wind turbine nonlinear system. These measured values are important both for the controller performance as well as the supervisory module of the wind turbine. The proposed control accommodation scheme is based on identified nonlinear filters, which can be also exploited to diagnose these faults. In particular, a nonlinear dynamic system estimation scheme is proposed to obtain these disturbance decoupled nonlinear filters, which provide the on-line estimate of the fault signals. The scheme is tested on a high-fidelity wind turbine simulator, in the presence of disturbance and measurement errors, along nominal operating conditions, including also different realistic fault situations. Tests on this simulated model show the the reliability features of the proposed active fault tolerant control scheme.