An Ultrahigh Frequency Partial Discharge Signal De-Noising Method Based on a Generalized S-Transform and Module...

Area of Science:

• Electrical Engineering
• Signal Processing
• Power Systems

Background:

• Partial discharge (PD) signals in power substations are crucial for high-voltage apparatus fault diagnosis.
• Electromagnetic interference introduces background noise into ultrahigh frequency (UHF) PD signals, hindering accurate analysis.
• Existing noise reduction methods may not sufficiently preserve signal integrity.

Purpose of the Study:

• To develop and validate a novel de-noising method for UHF PD signals.
• To effectively suppress both periodic narrowband noise and Gaussian white noise.
• To improve the accuracy of fault diagnosis and localization in high-voltage equipment.

Main Methods:

• Utilized the generalized S-transform and module time-frequency matrix for noise suppression.
• Employed the sub-matrix maximum module value method for periodic narrowband noise identification and cancellation.
• Applied singular value decomposition (SVD) with fuzzy c-means clustering for Gaussian white noise removal and signal recovery.

Main Results:

• The proposed method effectively suppressed background noise in both simulated and field-detected UHF PD signals.
• Achieved a higher signal-to-noise ratio (SNR) compared to four conventional de-noising techniques.
• Demonstrated reduced waveform distortion, preserving crucial signal characteristics for diagnosis.

Conclusions:

• The novel de-noising approach based on generalized S-transform and SVD is feasible and effective for UHF PD signals.
• The method offers superior performance in noise suppression and signal preservation.
• Enhanced PD signal quality facilitates more reliable high-voltage apparatus fault diagnosis and localization.