One of the main control methods of the surface layer of bearing parts is the eddy current testing method. Surface layer defects of bearing parts, like burns, cracks and some others, are reflected in the results of the rolling surfaces scan. The previously developed method for detecting defects from the image of the raceway was quite effective, but the processing algorithm is complicated and lasts for about 12 ... 16 s. The real non-stationary signals from an eddy current transducer (ECT) consist of short-time high-frequency and long-time low-frequency components, therefore a transformation is used for their analysis, which provides different windows for different frequencies. The wavelet transform meets these conditions. Based on aforesaid, a methodology for automatically detecting and recognizing local defects in bearing parts surface layer has been developed on the basis of wavelet analysis using integral estimates. Some of the defects are recognized by the amplitude component, otherwise an automatic transition to recognition by the phase component of information signals (IS) is carried out. The use of intelligent technologies in the manufacture of bearing parts will, firstly, significantly improve the quality of bearings, and secondly, significantly improve production efficiency by reducing (eliminating) rejections in the manufacture of products, increasing the period of normal operation of the technological equipment (inter-adjustment period), the implementation of the system of Flexible facilities maintenance, as well as reducing production costs.

The problem of recognition of defects in the surface layer of polished raceways of bearing rings according to the fractal dimension of the signals of the eddy current sensor is considered.