Sensor fault reconstruction for wind turbine benchmark model using a modified sliding mode observer

<p><span lang="EN-GB">This paper proposes a fault diagnosis scheme applied to wind turbine system. The technique used is based on modified sliding mode observer (SMO), which permits the reconstruction of actuator and sensor faults. A benchmark with real sequence speed exploited validate proposed detection scheme. Rotor speed, generator blade pitch angle, torque have different orders magnitude. As result, dedicated sensors are susceptible faults quite varying magnitudes, estimating simultaneous accuracy using classical SMO difficult. To address this issue, some modifications made classic SMO. In order test efficiency SMO, several scenarios been simulated, first in case separate then simulation results show that isolated, detected, reconstructed accurately faults, modification precisely reconstructed, even though they amplitudes; thus, relative gap does not exceed 0.08% for fault.</span></p>