Abstract:Empirical wavelet transform (EWT) is an adaptive tool for vibration signals processing and has been adopted for fault diagnosis of rolling bearings, while still suffers two weaknesses in detecting the weak, early-stage defects. First, adaptive and robust boundaries determination of the EWT segments is a challenge. Second, vestigial noise and unwanted vibrations still exist in the filtered signal, which may bury the weak fault signature and weaken the detection performance of early-stage faults. In this article, a two-step denoising strategy (TSDS) is proposed for early-stage fault detection, which includes frequency filtering and time-domain denoising steps. In the scheme, an automated EWT segmentation method is proposed for identifying each primary harmonic component; then, EWT-based correlated kurtogram is adopted to optimize the filtering frequency band such that to enhance the signal-to-noise ratio (SNR) of the fault information. Furthermore, the two-layer sliding-correlated kurtosis (TLSCK) algorithm is applied as the second step, which eliminates the vestigial noise and outputs pure periodic pulses, indicating the occurrence moments of the fault impulses. Thus, the weak, early-stage fault signature can be significantly enhanced and detected. A simulation study is conducted and validates the successfulness of the proposed scheme when the SNR is as low as -19 dB. A run-to-failure experiment verifies the effectiveness of the proposed method for weak faults identification, and comparison studies show that the proposed scheme could diagnosis bearing defects at much earlier moment than traditional methods like spectral kurtosis and dyadic wavelet transform.

Denoising by Wavelet Soft-Thresholding in Fault Analysis

Adaptive Wavelet Thresholding Based Ultrasonic Signal Denoising

The Application of De-noise Based on Compromise Algorithm Between Soft-thresholding and Hard-thresholding of Wavelet Coefficient for Fault Location

Signal Denoising Using Undecimated Second Generation Wavelet Transform

Research on an Adaptive Wavelet Denoising Method

A data-driven threshold for wavelet sliding window denoising in mechanical fault detection

New denoising method based on dual-tree complex wavelet transform and nonlinear time series

Application of Multiwavelet Adaptive Threshold Denoising in Fault Diagnosis

A Two-Step Denoising Strategy for Early-Stage Fault Diagnosis of Rolling Bearings

Denoising Analysis of Improved Soft Threshold Method

Research on a Novel Denoising Method of Ultrasonic Measurement Signal

Improvement of the Wavelet Threshold Filtering Algorithm Applied to Audio Signals

Adaptive Denoising by Soft- Thresholding with Wavelets

Investigation of microseismic signal denoising using an improved wavelet adaptive thresholding method

Multiple-analysis of Wavelet Transform and Selecting Threshold Value in the Applicant of One-dimension Signal Denoising

Application of an Improved Wavelet Threshold Denoising Method for Vibration Signal Processing

An Adaptive Wavelet Multilevel Soft Threshold Algorithm for Denoising Partial Discharge Signals

Application of Wavelet Analysis in Denoising Seismic Data

An improved wavelet threshold denoising method and its application

Wavelet De-noising Algorithm and Application Based on Improved Threshold Function

A New Wavelet Threshold Function and Denoising Application