Detection of Bearing Defects Based on Vibration Signature Analysis and DWT

This paper deals with the detection and diagnosis of mechanical defects and more particularly those found in the ball bearings that can equip an electric machine. Generally, the defects in the bearings are diagnosed through the search for the characteristic frequencies associated with the components of the bearing. These frequencies are supposed to be known a priori or estimated. This paper proposes to discriminate between bearing defects depending on the location and the severity of the defect. The method relies on the extraction of the frequency parameters representative of the overall frequency signature of the defects in order to make the classification. For our approach, the analysis of the experimental results shows that the Discrete Wavelet Transform (DWT) applied to these parameters makes it possible to discriminate with precision the different types of defects.