Detection of partial discharge patterns in hybrid high voltage power transmission lines based on parallel recognition...

Area of Science:

• Electrical Engineering
• Signal Processing
• Artificial Intelligence

Background:

• Hybrid high-voltage power transmission lines (HHVPTLs) present unique challenges for condition monitoring due to their complex structures.
• Existing partial discharge (PD) detection algorithms for HHVPTLs suffer from low efficiency, poor accuracy, and lack of parallel analysis capabilities.

Purpose of the Study:

• To propose a novel parallel recognition method for partial discharge patterns in HHVPTLs.
• To address the limitations of current PD detection algorithms by improving efficiency and accuracy.

Main Methods:

• Wavelet packet decomposition was used to process non-smooth PD signals and form attribute vectors.
• An improved backpropagation neural network (BPNN), optimized by a beetle optimization (DBO) algorithm with orthogonal contrastive learning (IDBO-BPNN), was employed for pattern recognition.
• The IDBO-BPNN model was validated using PD data from a Kaggle competition.

Main Results:

• The proposed IDBO-BPNN model achieved a recognition accuracy of 95.5% for PD detection in HHVPTLs.
• The model demonstrated superior performance compared to DBO-BPNN and other recognition algorithms, with a 5.33% accuracy improvement over DBO-BPNN.
• The method accurately identifies the occurrence and phase of PDs, enhancing overall efficiency and practical applicability.

Conclusions:

• The IDBO-BPNN model offers a highly accurate and efficient solution for parallel partial discharge pattern recognition in HHVPTLs.
• This advanced method significantly improves the condition monitoring capabilities of hybrid power transmission infrastructure.
• The study highlights the potential of integrating advanced signal processing and optimized neural networks for critical power system diagnostics.