Robust fault estimation for wind turbine pitch and drive train systems

Abstract

The reliability and accuracy of the wind conversion system largely depend on the early detection and diagnosis of faults. In this paper, a novel fault estimator for wind turbine pitch and drive train systems is developed. The main objective is to estimate actuator and sensor faults along with the system states while mitigating the impact of process disturbances and noises. To accomplish this, an augmented state is created by combining the states of the system and different faults. Subsequently, an Unknown Input Observer (UIO) is developed to estimate them simultaneously. The UIO matrices are obtained by optimizing a multi-objective function formed by transforming states and faults estimation errors into the frequency domain using a genetic algorithm. Compared with other approaches, particularly H∞, the proposed technique shows great superiority in accurately estimating various actuators and sensors faults.