A Novel Hybrid Approach for Partial Discharge Signal Detection Based on Complete Ensemble Empirical Mode...

Area of Science:

• Electrical Engineering
• Signal Processing
• Noise Reduction

Background:

• Partial discharge (PD) detection is crucial for electrical insulation monitoring.
• White noise significantly interferes with accurate PD signal analysis.
• Existing methods like EMD and EEMD face challenges with mode mixing and noise suppression.

Purpose of the Study:

• To develop a novel noise-elimination method for PD detection.
• To enhance the accuracy of PD signal extraction in noisy environments.
• To improve upon existing decomposition techniques for PD analysis.

Main Methods:

• Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) for signal decomposition.
• Approximate Entropy (ApEn) for identifying and eliminating noisy intrinsic mode functions (IMFs).
• Correlation coefficient analysis for selecting dominant PD features and reconstructing the signal.

Main Results:

• CEEMDAN effectively decomposes signals into IMFs with distinct frequency scales.
• ApEn successfully filters out noise-corrupted IMFs.
• The proposed CEEMDAN-ApEn method demonstrates superior noise suppression and PD pulse extraction compared to EMD and EEMD.
• Improved reconstruction accuracy and reduced iteration numbers were observed.

Conclusions:

• The CEEMDAN-ApEn fusion algorithm offers a robust solution for noise reduction in PD detection.
• This method effectively extracts characteristic PD pulses, enhancing diagnostic capabilities.
• The approach shows significant potential for both simulated and on-site PD signal analysis.