ELECTROMAGNETIC TORQUE ESTIMATION BY KALMAN FILTER FOR DYNAMIC ECCENTRICITY FAULT DETECTION

Abstract

In this paper, the spectral analysis of the electromagnetic torque is used to detect dynamic eccentricity faults on induction machine. This faults lead to a specific signature on the torque spectrum related to the mechanical rotation pulsation. The frequency of this component depends on the slip and its amplitude increases with the load level. Instead of using an expensive torque measurement, the electromagnetic torque is estimated by a Kalman filter using current, voltage and speed measurements. The estimated torque and its spectrum are compared to the measured ones in order to highlight the performances of the Kalman filter. Using an off-line Fast Fourier Transform (FFT) analysis, experimental results show that the presence of dynamic eccentricity is detectable by monitoring the specific harmonic on the estimated electromagnetic torque spectrum.