Vibration Signal Denoising Method Based on ICFO-SVMD and Improved Wavelet Thresholding.

Area of Science:

• Mechanical Engineering
• Signal Processing
• Computational Intelligence

Background:

• Rotating machinery generates non-stationary, multi-component vibration signals.
• Strong background noise often masks critical fault features, reducing diagnostic accuracy.
• Existing denoising methods struggle with complex noise interference.

Purpose of the Study:

• To develop an advanced joint denoising method for contaminated vibration signals.
• To enhance the reliability of fault diagnosis in rotating machinery.
• To improve the detection of weak fault signatures.

Main Methods:

• A hybrid approach combining improved cordyceps fungus optimization algorithm (ICFO), successive variational mode decomposition (SVMD), and wavelet thresholding.
• ICFO adaptively optimizes SVMD parameters using enhanced initialization and mutation strategies.
• Signal decomposition, component classification using Pearson correlation, and adaptive wavelet thresholding for reconstruction.

Main Results:

• The proposed method significantly improved signal-to-noise ratio (SNR) and reduced root mean square error (RMSE) across various noise levels.
• Achieved superior performance compared to VMD, SVMD, VMD-WTD, CFO-SVMD, and WTD in simulations.
• Demonstrated effectiveness on real-world bearing and gearbox data, preserving fault frequencies with low residual variance.

Conclusions:

• The joint denoising method offers a robust solution for analyzing noisy vibration signals in rotating machinery.
• It effectively suppresses noise while retaining essential fault information for accurate diagnostics.
• This technique enhances the potential for early fault detection and predictive maintenance.