Partial Discharge Fault Diagnosis Based on Multi-Scale Dispersion Entropy and a Hypersphere Multiclass Support Vector...

Area of Science:

• Electrical Engineering
• Materials Science
• Signal Processing

Background:

• Partial discharge (PD) analysis is crucial for diagnosing insulation health in electrical equipment.
• Traditional PD fault diagnosis methods face limitations in accuracy and effectiveness.
• Developing advanced feature extraction techniques is essential for improving PD analysis.

Purpose of the Study:

• To propose a novel feature extraction approach for PD fault analysis.
• To enhance the accuracy of PD pattern recognition in electrical equipment.
• To overcome the limitations of conventional PD diagnostic tools.

Main Methods:

• Variational Mode Decomposition (VMD) was used to decompose PD signals into intrinsic mode functions (IMFs).
• Multi-scale dispersion entropy (MDE) was calculated for selected IMFs.
• Principal Component Analysis (PCA) extracted key features from MDE values.
• A hypersphere multiclass support vector machine (HMSVM) classified PD patterns using the extracted features.

Main Results:

• The VMD-MDE approach effectively extracts dominant PD features.
• The proposed method demonstrates superior performance in PD fault diagnosis.
• Experiment results validate the effectiveness of the VMD-MDE feature extraction technique.

Conclusions:

• The VMD-MDE combined with HMSVM offers a powerful and effective solution for PD fault diagnosis.
• This novel approach significantly improves the accuracy and reliability of insulation condition assessment.
• The study highlights the potential of advanced signal processing techniques in electrical equipment maintenance.