Banks of estimators and decision mechanisms for pitch actuator and sensor FE in wind turbines

Abstract

Wind turbines are prone to multiple different faults and input observability conditions are not always guaranteed for these faults. In such cases, it is not possible to build estimators which provide appropriate fault estimates for its further use in active FTC schemes such as fault tolerant MPC. Provided that these faults are generally non-simultaneous, we make use of this property for building banks of model-based estimators and statistical-based decision mechanisms that provide appropriate fault estimates for enhancing active FTC capabilities. We apply these strategies to a well-known wind turbine FDI and FTC benchmark and we show the effectiveness of the bank of estimators and decision mechanisms for estimating the faults occurring in the pitch system of a wind turbine.