Application of advanced signal separation methods in a complex case of a bearing fault

The vibration signatures of machinery are widely used for mechanical diagnostics. The signals excited by the rotating components are transmitted to a sensor through the machine structure. Therefore, the measured signal contains the signals generated by the rotating components and filtered by the structure transfer function.

In some cases, it may be quite difficult to isolate the signals that are related to the defective component. This is especially true in cases of defective bearings. The difficulties are due to the facts that the signals related to the defect may be weak, the signals may be masked by other strong vibration sources (for example gears, shafts, pumps or other defects), the generated signals may be spread over a large range of frequencies and there may be a strong effect from the structure response.

This paper presents and illustrates the application of advanced signal separation methods in a complex case of a bearing fault. The signal separation methods include algorithms for removing the synchronous parts of the signal and algorithms for removing the effects of the structure response. The presented case demonstrates the power and efficiency of the methods to isolate and highlight the signals associated with the defect.