文档介绍：Proceedings of ICSP ' 96
The Application of Wavelet Transform and Artificial works
in Machinery Fault Diagnosis
Wu Yousheng, Sun Qiao, Pan Xufeng, Li Xiaolei
(Department of Vehicle Engineering, Beijing Institute of Technology, 10008 1)
Abstract: Wavclct translbnn and Artilicial works the non-stationar) sharp-variation signals often caused
(Aims) arc briellv described. 'lhen both of them are applied bj soinc sudden faults, e g . the broken of gears or shafts
comprehensively to machinery fault diagnosis. Wavelet trans- Wavclct IS a new developing signal ssing means It
form is used to pre-process data and extract feature vectors is locali~edboth in time and frequency domains So it is
Anns is uscd to identify fault types. 13y Wavelet transform, the possible to characterize the local singularities of ma-
dimension of' the feature vector is greatly decreased aid the chinery vibration signals based on the coefficieiits m a
noiscs are rcrtrained as well. ~lliusthe construction of the A" wavelct orthonormal basis expansion. Hcrc wc use
is simplified and the calculation speed is raised \vithout lower- Wavelet transform to process vibration data picked up
1:or coInparing, two types of features are extracted. from a machine and extract features that arc regarded as
Such diagnosing Incasurc is proved to be cllicient by ail experi- an Anns' lcarning