An adaptive signal processing method for extraction of a weak bearing signal

Bearing is the most frequently and easily failed component in any rotating machine. The extraction of the bearing feature signal is critical for signal analysis and fault diagnosis of bearings. In this paper, an adaptive signal processing method is proposed to extract a weak bearing signal from a raw vibration signal. To provide sufficient extremes for signal decomposition using the ensemble empirical mode decomposition (EEMD) method, a spectral-kurtosis-based band-pass filter is firstly used to process the raw signal, and the EEMD method with parameter optimization is then used to further extract the weak bearing signal. Experimental results of a faulty bearing demonstrate the effectiveness of the proposed method.