Target Detection Strategy for Intelligent Inspection System in Substation Areas Based on YOLOv8 Algorithm
Year: 2025
Type: conference-paper
Abstract: To address the limitations of current object detection algorithms for intelligent substation equipment inspection—such as weak learning capability and model overfitting—this paper proposes an improved YOLOv8-based model to enhance the accuracy of defect detection in substations. Firstly, a Frequency-Domain Feature Fusion Module (FDFFM) is integrated to extract and effectively fuse key information from the texture features of power equipment defects, thereby enhancing the model’s ability to recognize small defects under complex backgrounds with greater robustness. Secondly, a Composite Coordinate Attention Module (CCAM) is introduced to reinforce the model’s feature learning capacity for substation equipment defects, significantly improving the detection of small targets and low-contrast defects. Finally, the improved YOLOv8 algorithm is compared with multiple object detection algorithms on the same equipment defect dataset to validate its accuracy. Experimental results show that, compared to a similarly enhanced YOLOv7 algorithm, the proposed YOLOv8 model achieves a 5.12% increase in average accuracy for detecting substation instrument defects. These results demonstrate that the proposed method fully meets the requirements of defect detection in intelligent inspection systems for substation areas.
Authors: Yang Liu, Zhang Mingyuan, Ji Yulei
Institution: State Grid Corporation of China (China)
Language: English
Topic: Power Line Inspection Robots
Subfield: Mechanical Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4416321040