Broken-bar rotor fault detection in squirrel-cage induction motors at presence of sensor faults using adaptive Unscented Kalman filter

Abstract

In this paper, adaptive Unscented Kalman filter will be used for robust fault detection of a broken bar rotor in a squirrel-cage induction motor. In induction motors, broken bar rotor fault, present itself by increasing rotor resistance. Therefore, induction motor model is developed and rotor resistance is considered as one of the system states. Then General and Adaptive form of Unscented Kalman filter is applied to estimate model states by considering the nonlinear plant model. Finally, it will be shown that, at presence of errors such as offset or abnormal sensor measurements, induction motor states can be estimated by adaptive unscented kalman filter more accurately than by general unscented kalman filter. Therefore, fault detection of broken-bar rotor is performed more reliably.