TY - JOUR
T1 - Extracting repetitive transients for rotating machinery diagnosis using multiscale clustered grey infogram
AU - Li, Chuan
AU - Cabrera, Diego
AU - De Oliveira, José Valente
AU - Sanchez, René Vinicio
AU - Cerrada, Mariela
AU - Zurita, Grover
N1 - Publisher Copyright:
© 2016 Elsevier Ltd. All rights reserved.
PY - 2016/8/1
Y1 - 2016/8/1
N2 - Local faults of rotating machinery usually result in repetitive transients whose impulsiveness or cyclostationarity can be employed as faulty signatures. However, to simultaneously accommodate the impulsiveness and the cyclostationarity is a challenging task for rotating machinery diagnostics. Inspired by recently-reported infogram that is sensitive to either the impulsiveness or the cyclostationarity using spectral negentropy defined in time domain or frequency domain, a multiscale clustering grey infogram (MCGI) is proposed by combining both negentropies in a grey fashion using multiscale clustering. Fourier spectrum of the vibration signal is decomposed into multiple scales with different initial resolutions. In each scale, fine segments are grouped using hierarchical clustering. Meanwhile, both time-domain and frequency-domain spectral negentropies are taken into account to guide the clustering through grey evaluation of both negentropies. Numerical simulations and experimental tests are carried out for validating the proposed MCGI. For comparison, peer methods are applied to challenge different noises and interferences. The results show that, thanks to the multiscale clustering of the spectrum and the grey evaluation of both negentropies, the present MCGI is robust in extracting the repetitive transients for the rotating machinery diagnosis.
AB - Local faults of rotating machinery usually result in repetitive transients whose impulsiveness or cyclostationarity can be employed as faulty signatures. However, to simultaneously accommodate the impulsiveness and the cyclostationarity is a challenging task for rotating machinery diagnostics. Inspired by recently-reported infogram that is sensitive to either the impulsiveness or the cyclostationarity using spectral negentropy defined in time domain or frequency domain, a multiscale clustering grey infogram (MCGI) is proposed by combining both negentropies in a grey fashion using multiscale clustering. Fourier spectrum of the vibration signal is decomposed into multiple scales with different initial resolutions. In each scale, fine segments are grouped using hierarchical clustering. Meanwhile, both time-domain and frequency-domain spectral negentropies are taken into account to guide the clustering through grey evaluation of both negentropies. Numerical simulations and experimental tests are carried out for validating the proposed MCGI. For comparison, peer methods are applied to challenge different noises and interferences. The results show that, thanks to the multiscale clustering of the spectrum and the grey evaluation of both negentropies, the present MCGI is robust in extracting the repetitive transients for the rotating machinery diagnosis.
KW - Fault diagnosis
KW - Multiscale clustered grey infogram
KW - Repetitive transient
KW - Rotating machinery
KW - Spectral negentropy
UR - http://www.scopus.com/inward/record.url?scp=84977955865&partnerID=8YFLogxK
U2 - 10.1016/j.ymssp.2016.02.064
DO - 10.1016/j.ymssp.2016.02.064
M3 - Article
AN - SCOPUS:84977955865
SN - 0888-3270
VL - 76-77
SP - 157
EP - 173
JO - Mechanical Systems and Signal Processing
JF - Mechanical Systems and Signal Processing
ER -