Abstract:This paper presents a new methodology for the diagnosis of induction machine faults. A modern and efficient diagnosis technique should be non-destructive. The proposed technique is based on the vibratory behavior analysis.

Introduction:Vibration analysis is a very interesting and recent technique because any change in the mechanical or electrical conditions of the machine affects its dynamic conditions and thus its vibration behavior.

Methodology:The virtual works method is applied locally for the magnetic force computation. To calculate the dynamic response of the stator to the magnetic force stresses, the dynamic equation must be solved. The vibratory response (acceleration) of the stator is equal to a linear combination of associated mode shapes.

Numerical calculation codes based on finite elements method are developed under Matlab environment to diagnosing rotor bar breakage, short circuit and other faults. These codes were applied in healthy and faulty cases of the machine.

The Fast Fourier Transform was investigated. So, we compared between the frequency spectra of the accelerations in the faulty and healthy cases, to predict the defect.

Results:The obtained results prove that vibration analysis technique has the advantage of allowing the detection and the localization of defects. More details and results proving the interest of this technique for the monitoring of the induction machines will be presented in the full paper.

Conclusion:This study has given more understanding of the dependent roles of vibration analysis in predicting and diagnosing machine faults. So, we have proved the interest of this technique for the monitoring of the electrical machines.