Cable online partial discharge detection and state evaluation based on deep belief network and swarm intelligence...

Area of Science:

• Electrical Engineering
• Power Systems
• Artificial Intelligence

Background:

• Urban power grids increasingly rely on high-voltage cable lines for transmission.
• Insulation degradation due to partial discharge (PD) poses a significant risk to grid reliability.
• Existing PD detection methods face limitations in sensitivity, real-time performance, and scalability.

Purpose of the Study:

• To develop an efficient and reliable online monitoring framework for cable PD detection and state assessment.
• To improve the accuracy, real-time performance, and state evaluation capabilities of PD detection systems.
• To provide robust technical support for the safe operation of urban power grid cable assets.

Main Methods:

• Integration of an improved Deep Belief Network (DBN) with an Inertial Krill-Herd Algorithm with Differential Bat Adaptation (IKHA-DBA).
• Enhancement of the DBN model using DropConnect and elastic weight consolidation techniques.
• Optimization of the DBN using the IKHA-DBA algorithm for superior PD detection.

Main Results:

• The improved DBN model demonstrated significant gains in precision, recall, and AUC compared to ResNet50-SE on the B0005 dataset.
• The final IKHA-DBA optimized DBN model further improved detection accuracy and robustness on both B0005 and B0006 datasets.
• The proposed framework achieved comprehensive improvements in accuracy, robustness, and real-time performance for cable PD online detection.

Conclusions:

• The developed framework offers a reliable solution for online partial discharge detection in urban power grid cables.
• The integration of advanced AI algorithms significantly enhances the capabilities of existing PD monitoring technologies.
• This research provides crucial technical support for ensuring the safe and stable operation of critical power infrastructure.