Bearing fault indicator in induction machine using stator current spectral analysis

This paper deals with the application of motor current spectral analysis for the detection of rolling bearings damage in asynchronous machines. Vibration measurement is widely used to detect faulty operations. However, this approach is expensive and cannot always be performed. An alternative is to base the monitoring on electrical quantities such as the machine stator current which is often already measured for control and detection purposes. Firstly, to define an automatic detector for bearing faults based on mechanical quantities, vibration spectra are analyzed. Afterwards, it is shown that bearing faults induce mechanical load torque oscillations. Then, a theoretical stator current model in case of load torque oscillations demonstrates the presence of phase modulation. Related sideband components on current spectrum can be used for detection. Experimental measurements show that a transfer function including resonance links amplitudes of sideband components to the torque oscillation frequency. This singularity will be used to improve the detection efficiency. A fault detector using the energy of stator current in specific frequency ranges is then proposed. The efficiency of the detector is studied for different operating conditions. The supply frequency values are particulary investigated to demonstrate that using the resonance previously described could improve detection performances.