A New Denoising Method for UHF PD Signals Using Adaptive VMD and SSA-Based Shrinkage Method.

Area of Science:

• Electrical Engineering
• Signal Processing
• Materials Science

Background:

• Noise suppression is critical for the partial discharge (PD) ultra-high frequency (UHF) method in diagnosing high-voltage electrical equipment insulation defects.
• Existing denoising algorithms often fail to address multiple noise types simultaneously and neglect essential feature preservation.

Purpose of the Study:

• To propose a novel denoising method for ultra-high frequency (UHF) partial discharge (PD) signals.
• To address the limitations of existing methods by simultaneously reducing various noises and preserving signal features.

Main Methods:

• Utilized variational mode decomposition (VMD) with an automatic mode number selection for signal decomposition into band-limited intrinsic mode functions (BLIMFs).
• Employed a kurtosis-based rule to identify effective BLIMFs (eBLIMFs).
• Applied singular spectrum analysis (SSA)-based thresholding for residual white noise suppression in eBLIMFs, followed by signal reconstruction.

Main Results:

• The proposed method successfully decomposed UHF PD signals and effectively suppressed noise.
• Feature preservation was maintained during the denoising process.
• New evaluation indices demonstrated the method's superiority over traditional techniques in noise suppression and feature retention.

Conclusions:

• The developed denoising technique offers a significant improvement for UHF PD signal analysis.
• The method proves effective across simulated, laboratory, and field data, enhancing insulation defect diagnosis.
• This approach provides a robust solution for noise reduction while preserving critical signal characteristics.