Fault diagnosis of wind energy conversion systems with sensor faults

In this paper, a novel filter based fault detection and isolation strategy is presented for variable-speed wind energy conversion systems (WECSs) in the presence of sensor faults. Firstly, a nonlinear physical model of a WECS is built by integrating aerodynamic rotor, a drive train, a generator and a converter. The typical sensor faults are also modelled. Since the wind speed is usually a non-Gaussian signal, a novel filter is then presented for WECSs via improved minimum error entropy criterion. Based on the established filter, residual signals are generated, and accordingly the detection and isolation rule for the sensor faults is obtained by evaluating residual signals. Finally, the proposed fault detection and isolation scheme is applied to a WECS. Simulations results indicate the sensor faults can be detected and isolated effectively.