application of nonlinear kalman filters for Model-Based Fault Detection in Induction Motors

In this paper, a model-based fault detection method for induction Motors is presented. A new filtering technique based on Unscented/Extended Kalman filters, is utilized as a state estimation tool in broken rotor bars detection of induction motors. Failure events are detected by jumps in the estimated parameters of model. We use the UKF/EKF to estimate the value of the rotor resistance. Using the merits of these recent nonlinear estimation tools, rotor resistance of an induction motor is estimated only by the sensed stator currents and voltages information. In order to compare the estimation performances of the EKF and UKF, both observers are designed for the same motor model and run with the same covariance matrices under the same conditions. The results show the superiorly of UKF over EKF in highly nonlinear systems, as it provides better estimates of which is most critical for rotor fault detection.