Research on insulator defect detection algorithm of transmission line based on CenterNet

Area of Science:

• Electrical Engineering
• Computer Vision
• Artificial Intelligence

Background:

• Insulator reliability is crucial for stable electric power system operation.
• Defective insulator detection is a key challenge in smart grid systems.
• Traditional methods suffer from low accuracy and poor real-time performance.

Purpose of the Study:

• To develop a highly accurate and efficient insulator defect detection method.
• To improve the real-time detection capabilities for smart grid applications.
• To address limitations of existing transmission line inspection techniques.

Main Methods:

• Utilized a simplified backbone network within the CenterNet framework.
• Incorporated an attention mechanism to enhance detection accuracy by filtering irrelevant information.
• Employed super-resolution reconstruction for preprocessing blurred images to augment the dataset.

Main Results:

• Achieved an Average Precision (AP) of 96.16%.
• Reached a reasoning speed of 30 frames per second (FPS) on NVIDIA GTX 1080.
• Demonstrated significant improvements in detection accuracy compared to Faster R-CNN, YOLOV3, RetinaNet, and FSAF.

Conclusions:

• The proposed CenterNet-based method significantly enhances insulator defect detection accuracy and efficiency.
• The integration of attention mechanisms and super-resolution reconstruction proves effective.
• The method offers a viable solution for real-time monitoring in smart grid systems.