Presentation Details
  1. Home
  2. Presentation List
  3. Presentation Details
WLGD-Net: A Wavelet Local–Global Network for Denoising Rolling Bearing Vibration Signals
ID：94 View Protection：ATTENDEE Updated Time：2026-09-24 21:45:25 Hits：11 Poster Presentation

Start Time：Pending（Asia/Shanghai）

Duration：Pending

Session：No Session »

No files

Abstract
Fault-related transients in rolling bearing vibration signals are often masked by mixed disturbances, while clean reference measurements are rarely available. This paper proposes a Wavelet Local–Global Denoising Network (WLGD-Net), which combines fixed five-level Haar decomposition, local convolutional and self-attention branches, and residual subtraction. The network is trained using physically simulated mechanical signals with composite disturbances and requires no fault-frequency input during inference. A simulation with a known clean reference and a laboratory outer-race-fault record are used to evaluate waveform recovery, transient preservation, fault-harmonic retention, and residual leakage. Results show that WLGD-Net suppresses mixed disturbances while retaining recognizable fault transients and envelope-spectrum harmonics under the tested conditions.
Keywords
vibration denoising; wavelet transform; residual learning; envelope spectrum; physical simulation
Speaker
Yangkun Li
Mr. North China Electric Power University

Submission Author
Yangkun Li North China Electric Power University
Liqi Zhao North China Electric Power University
Wenwei Luo North China Electric Power University
Jiatie Li North China Electric Power University
Aijun Hu North China Electric Power University
Submit Comment
Verify Code Change Another
All Comments
Important Date

  • Conference Date

    Nov 06

    2026

    to

    Nov 08

    2026

  • Oct 15 2026

    Draft paper submission deadline

Sponsored By
IEEE Instrumentation and Measurement Society
Organized By
Sichuan University
Contact Information
Login to view full contact details
Previous Conferences