Abstract
The periodic impulse characteristics caused by rolling bearing damage are weak in the incipient failure stage. Thus, these characteristics are always drowned out by background noise and other harmonic interference. A novel approach based on multi-resolution singular value decomposition (MRSVD) is proposed in order to extract the periodic impulse characteristics for incipient fault detection. With the MRSVD method, the vibration signal is first decomposed to obtain a group of approximate signals and detailed signals with different resolutions. The first detail signal is mainly composed of noise and the last approximate signal is mainly composed of harmonic interference. Combined with the kurtosis index, the hidden periodic impulse signal will be extracted from the detail signals (in addition to the first detail signal). Thus, the incipient fault detection of a rolling bearing can be fulfilled according to the envelope demodulation spectrum of the extracted periodic impulse signal. The effectiveness of the proposed method has been demonstrated with both simulation and experimental analyses.
Highlights
When a rolling bearing has a local injury in addition to its periodic rotation, the damaged surface will be in contact with other parts that can produce periodic pulse impact, and arouse the natural frequency vibration of the inner and outer rings [1,2]
Supposing that the the periodic component by the fault, the harmonic generated by the vibration gatheredshock vibration signal ofcaused the rolling bearing is x(i), x(i)component can be expressed as Formula of the equipment or the harmonic current, and the background noise component
A new technique based on multi-resolution singular value decomposition (MRSVD) has been developed for the initial fault detection of a rolling bearing
Summary
When a rolling bearing has a local injury in addition to its periodic rotation, the damaged surface will be in contact with other parts that can produce periodic pulse impact, and arouse the natural frequency vibration of the inner and outer rings [1,2]. The purpose of the preprocessing is to remove the noise and harmonic interference and extract the periodic shock vibration components that are related to the fault for demodulation analysis. In contrast to the above frequency-separation-based methods, a new SVD-based preprocessing method, that is, multi-resolution singular value decomposition (MRSVD) [46], is introduced in this study with the objective of extracting the periodic shock vibration from the raw vibration signal which is acquired at the incipient failure stage of a rolling bearing. The MRSVD method has a good application prospect for the initial fault detection of a rolling bearing
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
