In the previous work of authors, the authors have presented an automatic fault feature extraction method, called ensemble superwavelet transform (ESW), based on the combination of tunable Q-factor wavelet transform (TQWT) and Hilbert transform. However, the nonstationary fault feature ratio which defined to guide the optimal wavelet basis selection does not take the interferences of high-frequency components into consideration. In addition, the original ESW utilizes one optimal subband to reconstruct the signal, which may result in the leakage of useful fault features. The present paper improves the ESW to address these problems. Specifically, the authors modify the definition of fault feature ratio by eliminating the high-frequency components when calculating total amplitudes of Hilbert envelope spectrum. Moreover, for the purpose of preserving more useful fault features and recovering the signal more accurately, a novel approach to reconstruct the processed result by incorporating two optimal subbands is proposed in this paper. The comprehensive comparisons by processing two simulation signals are provided to verify the effectiveness and utility of the improved ESW. Moreover, the improved ESW is applied to a range of engineering applications, and the obtained results demonstrate that the improved ESW can act as an effective technique in extracting weak fault features.
Improved Ensemble Superwavelet Transform for Vibration-Based Machinery Fault Diagnosis
Xi'an Jiaotong University,
Xi'an 710049, China;
New York University,
Brooklyn, NY 11201
Manuscript received July 11, 2015; final manuscript received December 30, 2015; published online March 15, 2016. Assoc. Editor: Laine Mears.
- Views Icon Views
- Share Icon Share
- Search Site
He, W., Zi, Y., Wan, Z., and Chen, B. (March 15, 2016). "Improved Ensemble Superwavelet Transform for Vibration-Based Machinery Fault Diagnosis." ASME. J. Manuf. Sci. Eng. July 2016; 138(7): 071012. https://doi.org/10.1115/1.4032568
Download citation file: