Foreign Object Detection System for Transmission Lines Based on NPU Chip ARM Platform

CN116343130BActive Publication Date: 2026-03-13YUNMENG POWER SUPPLY CO OF STATE GRID HUBEI ELECTRIC POWER CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-04-04
Publication Date
2026-03-13

AI Technical Summary

Technical Problem

Existing methods for identifying bird nests on power lines mainly employ techniques such as Hough transform line detection and background subtraction. However, the fixed camera angle limits bird nest identification, and manual inspections are inefficient and pose safety hazards.

Method used

A UAV-borne bird nest target recognition system based on an NPU chip ARM platform is adopted. The system uses the SSD-DenseNet neural network structure to detect bird nests on power lines and achieves real-time detection through data augmentation and model compression quantization.

Benefits of technology

This improved the accuracy and efficiency of bird nest target identification, enabling real-time, precise, and safe power line inspection by drones, avoiding power line failures caused by bird nests, and improving the reliability and safety of the power lines.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116343130B_ABST
    Figure CN116343130B_ABST
Patent Text Reader

Abstract

This invention relates to the field of foreign object identification and detection technology for power transmission lines based on the NPU chip ARM platform, and discloses a foreign object identification and detection system for power transmission lines based on the NPU chip ARM platform, including a UAV-borne bird nest target recognition system based on the NPU chip ARM platform. The UAV-borne bird nest target recognition system based on the NPU chip ARM platform includes a data augmentation module, the output signal of which is connected to a model quantization training module. A UAV-borne power line bird nest detection and recognition system based on the ARM platform is proposed. The neural network uses the SSD-DenseNet network model. By superimposing and enhancing the samples, the generalization ability of the neural network is improved, and the accuracy of bird nest target recognition is significantly improved. By integrating the model into the ARM platform, the UAV equipped with the ARM platform completes the task of real-time bird nest target detection. Through comparative experiments, it was finally determined that when the compression ratio of the SSD-DenseNet model is 0.3, the requirement for real-time bird nest target detection can be met on the ARM side.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Power transmission line fault real-time detection method based on edge calculation

    CN115311558A

  • Power transmission line foreign matter detection method based on improved YOLOv4

    CN115410087A