Multi-scale Mode Denoising (MSMD) Method for Weak Gear Fault Feature Extraction

Features caused by the gear fault are often submerged by the inference of the strong noise, which will lead to difficulties for the feature identification of the weak fault at the early stage. In addition, the desired weak feature would be modulated into several frequency bands, and then weaken the effect of the fault information identification. Motived by this issue, a MSMD method to extract and enhance the early weak fault signals is proposed in this paper. In this method, the collected raw vibration signals are decomposed into a series of band-limited intrinsic mode functions by variational mode decomposition, and then the sensitive modes including rich periodic amplitude modulation information are selected via proposed syncretic modulated impact index. The denoised signals can then be obtained through reconstructing the sub-signals generated from the sensitive modes using minimum entropy deconvolution. Finally, the fault feature can be identified clearly in the envelope spectrum of denoised signals. The results of simulation and experiment illustrate that the proposed method can effectively extract and enhance the weak fault features of gears.