变化注意力引导的变电站缺陷检测算法

Change Attention Guided Defect Detection Algorithm for Substations

摘要

变电站缺陷及时检测和处理是保障电力系统稳定运行的基础.针对现有变电站缺陷检测算法存在的背景干扰抑制不足、多尺度目标识别能力薄弱及双时相信息利用不充分等问题,本文提出一种变化注意力引导的变电站缺陷检测算法.该方法充分挖掘变电站双时相巡检图像中的时序差异信息,通过构建孪生骨干网络提取多尺度特征,并设计变化注意力引导模块,以双时相特征差异为注意力权重,动态增强缺陷区域的显著性,抑制背景干扰.同时,引入多尺度空洞特征提取模块与自适应特征融合结构,提升模型对不同尺度缺陷的检测能力.在自建的变电站双时相缺陷数据集上的实验结果表明,所提方法在平均精度均值、精确率、召回率等指标上显著优于Faster RCNN等主流检测模型.消融实验与热力图可视化分析进一步验证了变化注意力机制与特征融合策略的有效性,证明了该方法在复杂环境下具备良好的鲁棒性与工程应用潜力.

Abstract

Timely detection and handling of substation defects are fundamental to ensuring the stable operation of power systems.To address the issues in current substation defect detection algorithms,such as insufficient suppression of background interference,weak capability in recognizing multi-scale targets,and inadequate utilization of dual-temporal information,this paper proposes a change attention-guided substation defect detection algorithm.This method fully explores the temporal difference information in dual-temporal substation inspection images,extracts multi-scale features by constructing a Siamese backbone network,and designs a change attention guidance module.Taking the dual-temporal feature difference as attention weights,the module dynamically enhances the saliency of defect regions and suppresses background interference.Meanwhile,a multi-scale atrous feature extraction module and an adaptive feature fusion structure are introduced to improve the model’ s capability in detecting defects of varying scales.Experimental results on a self-built bi-temporal substation defect dataset demonstrate that the proposed method is significantly superior to mainstream detection models such as Faster RCNN in terms of mean Average Precision,Precision,Recall and other indicators.Ablation experiments and heatmap visualization analysis further validate the effectiveness of the change attention mechanism and feature fusion strategy,proving that this method has good robustness and engineering application potential in complex environments.

关键词

双时相图像 / 变电站巡检 / 变化注意力 / 变电站缺陷检测 / 自适应特征融合

Key words

dual-temporal images / substation inspection / change attention / substation defect detection / adaptive feature fusion

参考文献

[1] 罗悦冬, 杜思超, 李猷, 等. 基于深度特征感知的变电站设备典型缺陷检测方法[J/OL]. 南方电网技术, 2025 (2025-03-28).https://link.cnki.net/urlid/44.1643.TK.20250328.1131.003.

[2] LUO Y D, DU S C, LI Y, et al. Typical defect detection of substation equipment based on deep feature perception[J/OL]. Southern Power System Technology, 2025 (2025-03-28).https://link.cnki.net/urlid/44.1643.TK.20250328.1131.003.

[3] 白艳峰, 王立彪, 高卫东, 等. 面向电力设备缺陷检测的多模态层次化分类[J]. 中国图象图形学报, 2024, 29(7):2011-2023.

[4] BAI Y F, WANG L B, GAO W D, et al. Multi-modal hierarchical classification for power equipment defect detection[J]. Journal of Image and Graphics, 2024, 29(7):2011-2023.

[5] 秦佳鸣, 梁颖, 金继开. 变电站智能巡检系统实用化应用分析[J]. 电力与能源, 2024, 45(3):389-391.

[6] QIN J M, LIANG Y, JIN J K. Practical application analysis of intelligent substation inspection system[J]. Power & Energy, 2024, 45(3):389-391.

[7] 齐冬莲, 韩译锋, 周自强, 等. 基于视频图像的输变电设备外部缺陷检测技术及其应用现状[J]. 电子与信息学报, 2022, 44(11):3709-3720.

[8] QI D L, HAN Y F, ZHOU Z Q, et al. Review of defect detection technology of power equipment based on video images[J]. Journal of Electronics & Information Technology, 2022, 44(11):3709-3720.

[9] 赵振兵, 冯烁, 席悦, 等. 大模型时代:电力视觉技术新起点[J]. 高电压技术, 2024, 50(5):1813-1825.

[10] ZHAO Z B, FENG S, XI Y, et al. The era of large models:A new starting point for electric power vision technology[J]. High Voltage Engineering, 2024, 50(5):1813-1825.

[11] 陈鹏辉. 变电站场景下的图像变化检测技术研究[D]. 杭州: 浙江大学, 2021.

[12] CHEN P H. Research on image change detection technology in substation scene[D]. Hangzhou: Zhejiang University, 2021.

[13] REN S Q, HE K M, GIRSHICK R, et al. Faster R-CNN:Towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6):1137-1149.

[14] YING Y, WANG Y Z, YAN Y F, et al. An improved defect detection method for substation equipment[C]// 2020 39th Chinese Control Conference (CCC),27-29 July 2020, Shenyang,China.IEEE,2020:6318-6323.

[15] 张铭泉, 邢福德, 刘冬. 基于改进Faster R-CNN的变电站设备外部缺陷检测[J]. 智能系统学报, 2024, 19(2):290-298.

[16] ZHANG M Q, XING F D, LIU D. External defect detection of transformer substation equipment based on improved Faster R-CNN[J]. CAAI Transactions on Intelligent Systems, 2024, 19(2):290-298.

[17] GE Z, LIU S T, WANG F, et al. YOLOX:Exceeding YOLO series in 2021[EB/OL]. 2021:arXiv: 2107.08430.https://arxiv.org/abs/2107.08430.

[18] DONG H, YUAN M, WANG S, et al. PHAM-YOLO:A parallel hybrid attention mechanism network for defect detection of meter in substation[J]. Sensors, 2023, 23(13):6052.

[19] YUAN J H, WANG F L, HU Q L, et al. The substation defect detection method based on FEFPN-YOLOv7[C]// 2024 International Conference on Image Processing,Computer Vision and Machine Learning (ICICML),22-24 November 2024, Shenzhen,China.IEEE,2024:419-423.

[20] CHEN H R X, SONG J, HAN C X, et al. ChangeMamba:Remote sensing change detection with spatiotemporal state space model[J]. IEEE Transactions on Geoscience and Remote Sensing, 2024, 62:1-20.

[21] FENG Y C, JIANG J W, XU H H, et al. Change detection on remote sensing images using dual-branch multilevel intertemporal network[J]. IEEE Transactions on Geoscience and Remote Sensing, 2023, 61:4401015.

[22] 王鹏, 赵春晖, 周华良, 等. 基于多时相巡检图像的变电设备抗干扰缺陷检测[J]. 控制与决策, 2024, 39(3):885-892.

[23] WANG P, ZHAO C H, ZHOU H L, et al. Anti-interference defect detection of substation equipment based on multi-temporal inspection images[J]. Control and Decision, 2024, 39(3):885-892.

[24] 赵振兵, 马迪雅, 石颖, 等. 基于改进YOLOX的变电站仪表外观缺陷检测算法[J]. 图学学报, 2023, 44(5):937-946.

[25] ZHAO Z B, MA D Y, SHI Y, et al. Appearance defect detection algorithm of substation instrument based on improved YOLOX[J]. Journal of Graphics, 2023, 44(5):937-946.

[26] LINDENBERGER P, SARLIN P E, POLLEFEYS M. LightGlue:Local feature matching at light speed[C]// 2023 IEEE/CVF International Conference on Computer Vision (ICCV),1-6 October 2023, Paris,France.IEEE,2023:17581-17592.

[27] HU M, LI Y L, FANG L, et al. A²-FPN:Attention aggregation based feature pyramid network for instance segmentation[C]// 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR),20-25 June 2021, Nashville,TN,USA.IEEE,2021:15338-15347.

[28] NOMAN M, FIAZ M, CHOLAKKAL H, et al. ELGC-net:Efficient local-global context aggregation for remote sensing change detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2024, 62:1-11.

[29] YANG L X, ZHANG R Y, LI L D, et al. Simam: A simple, parameter-free attention module for convolutional neural networks[C]// Proceedings of the 38th International conference on machine learning. PMLR,2021:11863-11874.

[30] 李志财, 龙军, 钱育蓉, 等. 基于差异感知增强的遥感图像耕地变化检测网络[J]. 计算机工程与应用, 2026, 62(4):224-237.

[31] LI Z C, LONG J, QIAN Y R, et al. Difference-aware enhancement network for cropland change detection in remote sensing images[J]. Computer Engineering and Applications, 2026, 62(4):224-237.

[32] HE K M, ZHANG X Y, REN S Q, et al. Delving deep into rectifiers:Surpassing human-level performance on ImageNet classification[C]// 2015 IEEE International Conference on Computer Vision (ICCV),07-13 December 2015, Santiago,Chile.IEEE,2015:1026-1034.

[33] SOHAN M, THOTAKURA S R, REDDY V R. A review on YOLOv8 and its advancements[C]// International Conference on Data Intelligence and Cognitive Informatics.SingaporeSpringer Nature Singapore2024:529-545.

[34] WANG A, CHEN H, LIU L H, et al. YOLOv10:Real-time end-to-end object detection[C]// Advances in Neural Information Processing Systems 37,10-15 December 2024,Vancouver,BC,Canada.Neural Information Processing Systems Foundation,Inc.(NeurIPS), 2024:107984-108011.

[35] WANG C C, HE W, NIE Y, et al. Gold-YOLO:Efficient object detector via gather-and-distribute mechanism[C]// Advances in Neural Information Processing Systems 36,10-16 December 2023,New Orleans,Louisiana,USA.Neural Information Processing Systems Foundation,Inc.(NeurIPS), 2023:51094-51112.

[36] Anugerah Galang Persada, Igi Ardiyanto, Muhammad Bayu Sasongko, et al. A review of CAM-based visual explanation on diabetic retinopathy[J]. IEEE Access, 2026, 14:13737-13767.

基金资助

国家自然科学基金项目(62576132)