Unknown input observer design for faults estimation using linear parameter varying model. Application to wind turbine systems

Abstract

This paper proposes a sensor and actuator estimation algorithm based on linear parameter varying (LPV) model. Considering sensor noise and disturbance, an unknown input observer (UIO) is developed. In this scheme, by building an augmented system with a filter, sensor fault and noise of the original system become into a part of actuator fault and disturbance. According to this augmented system, an UIO and fault estimation method have been designed. Then, by solving the linear matrix equalities (LMEs) and the linear matrix inequalities (LMIs), the parameters of the UIO are obtained. In addition, we analyze the convergence of the observer. In order to verify the proposed method, a wind turbine system with torque actuator fault and pitch angle sensor fault has been tested. From simulation results, it presents an efficient performance on both state and fault estimation.