Abstract: In order to extract fault features of large-scale power equipment from strong background noise, a new method based on translation-invariant multiwavelets denoising using neighboring coefficients is proposed. Multiwavelets have several scaling functions and wavelet functions，possess the excellent properties that scalar wavelet cannot satisfy simultaneously, and match different characteristics of signals. Moreover, translation-invariant multiwavelets avoid Gibbs phenomena and their average process show superior denoising and maintain signal smoothness. Additionally, neighboring coefficient denoising considers relativity of coefficients and overcomes the deficiency of traditional threshold denoising. Therefore, neighboring coefficient denoising is introduced into translation-invariant multiwavelets and translation-invariant multiwavelets denoising using neighboring coefficients is applied to the diagnosis of a gear box and a locomotive rolling bearing. The diagnosis results of the gear box and the rolling bearing show that this method can effectively extract the fault feature of early gear crack, and the fault frequency of slight rub damage of bearing outer ring.

Key words: Monitoring and diagnosis, Multiwavelets, Neighboring coefficients, Signal denoising, Translation-invariant, Coupling of heat source, Energy partition, Form grinding, Grinding burn, Grinding temperature