<p>This paper presents a robust current sensor fault diagnosis method for Doubly Fed Induction Generator (DFIG). This method uses nonlinear estimators based on the extended Kalman filter (EKF) to observe the DFIG’s output current states. The diagnosis approach considers the correlation between rotor and stator currents in the DFIG model, recognising that changes in one current can affect other phase currents due to the machine’s interconnectedness. To isolate faulty sensors in the DFIG system, an effective method is applied. The EKF first estimates the currents and then compares them with sensor measurements to generate residuals. Next, a localisation block is employed to enhance the isolation process by learning about the complex relationships between sensor readings and fault occurrences in the DFIG system. The obtained simulation results demonstrate the effectiveness of the proposed method in detecting all scenarios of single and multiple current sensor faults in the <i>abc</i> natural reference frame.</p>