Robust Model-based fault diagnosis scheme for wind turbine system

This paper addresses the design of a robust sliding mode observer-based fault diagnosis scheme to reconstruct the sensor and actuator faults in a well-known validated wind turbine benchmark model. The method uses H_∞ concepts and introduces some LMIs to minimize the effect of uncertainty. The pitch subsystem parameter faults are transformed to the actuator like additive faults. The aerodynamic torque mapping, which is known as a look-up table, is approximated with an analytical form. Furthermore, in this paper some modifications are suggested to the observer discontinuous switching term to enable the estimation of the sensor and actuator faults, including simultaneous faults, with different order of magnitude. The simulation results show that the proposed modified observer faithfully reconstructs the sensor and actuator faults in the presence of uncertainty and the measurement noise.