Research on Signal Denoising of Pumped-Storage Units Based on Parameter-Adaptive VMD and Wavelet Thresholding

Area of Science:

• Engineering
• Signal Processing
• Mechanical Vibrations

Background:

• Vibration signals in pumped-storage units exhibit non-stationary and non-linear characteristics.
• These signals are often corrupted by strong background noise, hindering accurate analysis.
• Existing denoising methods may not adequately address these complex signal properties.

Purpose of the Study:

• To propose a robust joint signal denoising method for vibration signals from pumped-storage units.
• To enhance the signal-to-noise ratio (SNR) and reduce errors in collected data.
• To provide a reliable foundation for subsequent unit research and analysis.

Main Methods:

• Parameter-adaptive Variational Mode Decomposition (VMD) optimized using Improved Particle Swarm Optimization (IPSO).
• Cross-correlation analysis for screening decomposed signal components.
• Secondary denoising using wavelet thresholding on selected components.

Main Results:

• The proposed method effectively eliminates noise from complex vibration signals.
• Significant improvements in signal-to-noise ratio (SNR) were observed.
• A notable reduction in root mean square error (RMSE) was achieved compared to other methods.

Conclusions:

• The joint VMD and wavelet thresholding method offers superior denoising performance for pumped-storage unit vibrations.
• This approach provides a reliable data foundation for engineering applications.
• The study highlights substantial practical engineering significance for condition monitoring and fault diagnosis.