Bearing fault detection improvement using several stator current measurements in asynchronous drives

International audienceThis paper focuses on rolling bearings monitoring in induction motor, based on stator current measurements. Bearing defects induce slight load torque oscillations and then, specific fault sideband harmonics on stator currents. As stator current fault components are also of small amplitude, signal to noise ratio has to be considered. In order to implement an efficient automatic detection scheme of frequency fault components, based on energy extraction in the frequency domain, the signal to noise ratio of stator currents has to be increased. Using convolution theory, it is demonstrated that the use of several stator currents measurements allows to increase the signal to noise ratio and reduce the spectral estimation variance. Then, an intercorrelation of several stator currents is used to correlate spectral signatures. This method is assumed to be the most efficient one to improve the fault harmonics detection. An automatic detection scheme is proposed, based on the use of one or more stator current signals. Consequently, the advantages of taking into account more than one stator current measurements are analysed using simulation results and experimental measurements