1758. Detection and identification of mechanical faults by Kalman filtering in electric machines.

Kalman filtering technique is applied on vibration signals to detect eccentricity fault and bearing fault in induction motor. The vibration signals are acquired during load performance tests of healthy and faulty cases of induction motor. The state space equations needed for Kalman filter structure are constructed based on autoregressive model of the signals. The harmonics of fault related frequencies are identified from the spectrum of innovation signal, which is provided from the Kalman filter response, for vibration signal and also it is seen that Kalman filter reduce the noise level. [ABSTRACT FROM AUTHOR]