Prompt detection and isolation of faults is crucial in improving the reliability and safety of wind turbines and extending their lifespan. This thesis proposes a model?based fault detection and isolation scheme for the blade pitch positions of a wind turbine. An Unscented Kalman Filter (UKF), with a newly proposed de?correlation approach for process and measurement noises, is developed to simultaneously estimate the states and parameters of the WT's pitch position. Also a Fault Tolerant Control (FTC) approach has been applied to maintain the generator at its rated speed in the event of occurrence of pitch system faults. The FTC approach is formulated using the Fractional Order Terminal Sliding Mode Control (FO-TSMC) and relies on virtual actuators to retain the nominal pitching performance under faulty conditions. The proposed scheme was implemented to a 2.5 MW wind turbine and evaluated in the presence of several pitch actuator faults. Its performance was further compared to a PID controller and an Integer Order TSMC (IO_TSMC) approach.
