Research on Denoising Methods for Bearing Fault Vibration Signals Based on Wavelet Analysis and Envelope Demodulation

Abstract

Bearing vibration signals are often susceptible to environmental factors and measurement system interference during practical acquisition, resulting in low signal-to-noise ratios and difficulty in extracting fault features. To address this issue, this study investigates a denoising method for vibration signals based on wavelet analysis and envelope demodulation. First, the fundamental principles of wavelet threshold denoising and Hilbert envelope demodulation are introduced, and the influence of different thresholding strategies on denoising performance is analyzed. Subsequently, time-domain and frequency-domain analyses are performed on bearing acceleration vibration signals. The signals are processed using both wavelet threshold denoising and wavelet-based envelope demodulation denoising, and their denoising performances are comparatively evaluated. Results indicate that wavelet threshold denoising alone can suppress high-frequency noise to some extent, but residual noise remains. In contrast, combining envelope demodulation with subsequent wavelet denoising more effectively reduces high-frequency noise, enhances the rotational frequency and related characteristic frequency components of the bearing, and significantly improves the signal-to-noise ratio and feature distinguishability. The proposed method provides an effective signal preprocessing approach for bearing fault feature extraction and condition monitoring.