Abstract
Due to its practical importance, the diagnosis of rolling element bearing has attracted constant interest in the scientific community. At the incipient stage of a failure, the measured vibration signal typically consists of a series of repetitive transients immerged in background noise. Although they are usually carried in high frequency bands due to the high stiffness of bearings, they are fairly weak compared with surrounding noise and other interfering signals. In addition, taking random slips and fluctuations into account, the transients produced by impacts are not strictly periodic but rather tend to be random cyclostationary. This makes the diagnosis of rolling element bearing quite challenging and, consequently, various signal processing techniques have been developed for either the detection, the identification or the extraction of the fault, whose combination asks for a high level of expertise of the user. The aim of this paper is to address all these objectives at once, in the same algorithm, by proposing a semi-automated method that requires the setting of only one parameter. It is rooted on a probabilistic model, in the form of a mixture of Gaussians, endowed with a hidden variable that indicates the occurrence of impacts. The method is shown to be optimal for detection in the Neyman-Pearson sense, it returns an envelope spectrum comparable to the best that can be obtained by other means – which often require a careful pre-filtering step – from which fault frequencies can be identified, and it eventually returns the fault signal from which subsequent severity/health indicators can be computed. There is almost no demand on the user’s expertise (apart from setting the frequency resolution), even though the method does not address the decision part. The performance is investigated on synthetic signals and its robustness is also verified on several vibration signals measured on test-rigs. Results are found superior or at least equivalent to those of the conventional semi-automated method based on the fast kurtogram in combination with the envelope analysis.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.