Damage Assessment of Rolling Element Bearing Using Cyclostationary Processing of AE Signals with Electromagnetic Interference

Use of acoustic emissions (AE) has been shown to be of aid for bearing damage detection. Rolling element bearings with localized faults produce transient AE activity formed by bursts that repeat in an apparently periodic way. However, the produced signal is actually not periodic, but rather pure cyclostationary at the second order. Since AE are typically measured on the bearing housing, there is high attenuation of the waves, because of the dry metal–metal contact between the outer ring and the housing. In addition, AE instrumentation can be very sensitive to contamination by electromagnetic sources, such as frequency converters usually found near the measurement zones. Both situations combine and create adverse conditions for analysis. In this work a situation such as the described is presented. The measured AE is a corrupted signal where the valuable information is both weak and hidden. A filtering method by means of cyclostationary tools is presented and contrasted with other known methods. The proposed method is used subsequently for defect size estimation.