Sensor fault-tolerant control for a doubly fed induction generator in a smart grid

This study proposes a sensor fault-tolerant controller for a doubly fed induction generator connected to a smart grid. Mathematical models are the main tools for the synthesis of modern control systems; however, an accurate model for complex systems is not always available. Therefore, in this study, a recurrent high-order neural network trained with an extended Kalman filter is proposed to develop a mathematical model for a wind turbine with a doubly fed induction generator connected to a smart grid. The neural model is combined with a modified discrete-time sliding mode controller, which compensates for the presence of sensor faults on each of the state variables on both sides of the back-to-back converter. Real-time results are included to illustrate the effectiveness of the proposed scheme under five sensor faults.