Dynamic nonlinear unknown input observer for fault detection of induction motors

Abstract

Induction motors are widely adopted and can be found in many industrial plants. Early detection and diagnosis of faults for them are of great importance. In this paper, the model-based approach is used to detect the incipient faults as early as possible and to ensure the asymptotic convergence. The proposed method is used to detect changes in stator and rotor resistances which can be caused by stator winding short-circuit fault and broken rotor bars, respectively. To cope with the nonlinearities and at the same time load disturbances dynamic nonlinear unknown input observer (DNUIO) is used. In this paper, also a less conservative method is proposed to find Lipschitz constant for induction motors.