摘要
红外成像技术在变电设备监测中具有重要价值,能够通过温度异常识别潜在故障。然而,变电设备红外图像检测面临诸多挑战,如目标尺寸小,对比度低以及实时处理需求高,现有目标检测方法难以兼顾精度与效率。为此,提出一种基于改进YOLOv10n的轻量化检测模型,以提升变电设备在红外图像中的识别。首先,优化主干网络的C2f模块,引入全局到局部空间聚合模块,增强对小目标的聚焦能力并减少计算量。其次,在颈部网络中引入频率感知特征融合模块和频域感知路径聚合网络,有效解决目标边界模糊与偏移问题,并进一步减小模型体积。最后,采用SIoU损失函数替代传统的CIoU损失函数,提升目标定位精度并加速模型训练。基于变电站红外图像数据集进行验证,结果表明改进后的模型在检测性能上显著优于基线YOLOv10n模型。该模型的平均精度均值达到97.6%,较原始模型提升3.4%,同时计算复杂度和参数量大幅减少,适用于资源受限环境下的部署,对保障电网运行的安全性与稳定性具有重要意义。
Infrared imaging technology plays a crucial role in substation equipment monitoring by identifying potential faults through temperature anomalies.However,infrared image detection on substation equipment presents several challenges,such as small target size,low contrast,and high real-time processing demands.Existing target detection methods are difficult to balance accuracy and efficiency.To address these issues,this study proposes a lightweight detection model based on an improved YOLOv10n to enhance the recognition performance of substation equipment in infrared images.Firstly,the C2f module of the backbone network is optimized by introducing a global-to-local spatial aggregation module to improve focus on small targets while reducing computational load.Secondly,a frequency-aware feature fusion module and frequency-domain-aware path aggregation network are introduced into the neck network to effectively solve the problems of target boundary blurring and shifting,while further reducing the model size.Finally,the SIoU loss function is used to replace the traditional CIoU loss function,improving target location accuracy and accelerating model training.Experiments conducted on a substation infrared image dataset show that the improved model significantly outperforms the baseline YOLOv10n model in detection performance.Specifically,the model achieves a mean average precision of 97.6%,an improvement of 3.4%over the original model,while significantly reducing computational complexity and the number of parameters,making it suitable for deployment in resource-constrained environments.This has important implications for ensuring the safety and stability of power grid operations.
作者
潘国清
洪永学
邵天赐
杨强
PAN Guoqing;HONG Yongxue;SHAO Tianci;YANG Qiang(Guangzhou Power Supply Bureau of Guangdong Power Grid Co.,Ltd.,Guangzhou,Guangdong 510620,China;College of Electrical Engineering,Zhejiang University,Hangzhou,Zhejiang 310027,China)
出处
《广东电力》
北大核心
2025年第8期41-52,共12页
Guangdong Electric Power
基金
国家自然科学基金(52177119)。
