随着我国特高压远距离输电需求的不断增长，架空输电线路建设规模持续扩大。然而，西部地区输电线路多建设于梁塬地形中，输电线路绝缘子部件受到强风、雾霾等影响，出现破损、闪络等故障，从而影响输电安全。目前，绝缘子的智能巡检主要通过感知设备结合目标检测算法来实现。受复杂环境和资源受限的影响，导致获取的绝缘子图像质量无法满足智能检测的需求，影响了绝缘子检测精度。针对这一问题，本文从提升图像质量和改进检测算法两方面开展研究，以提升绝缘子的检测精度。主要工作如下：

（1）针对梁塬地形下，强风、雾霾等影响导致绝缘子出现破损、闪络的缺陷，建立了绝缘子缺陷数据集，并通过数据增强手段对数据集进行扩充。

（2）针对雾霾天气导致绝缘子图像细节丢失问题，提出了基于特征融合的循环生成对抗网络图像去雾算法。该算法在USID-Net算法中加入FCANet频率通道注意力，使网络更关注雾霾信息的频率特征，从而增强特征的表达能力。其次，针对USID-Net去雾图像存在色彩偏移现象，引入了颜色一致性损失，通过度量去雾图像与清晰图像在颜色空间的相似性，有效抑制颜色失真的问题。最后，采用DenseFuse算法将改进的USID-Net与AOD-Net去雾结果进行融合，提升图像的去雾效果。实验表明，相比USID-Net、AOD-Net，改进后算法PSNR值分别提升0.69dB、1.85 dB。

（3）因受供电、网络传输等资源限制，部署的前端感知设备分辨率不足以满足绝缘子智能检测的需求，提出了基于多尺度残差网络的超分辨率重建算法。该算法设计了多尺度特征注意力模块MSFA，并在MSFA中引入了RCCA与FPA注意力机制，引导模型精准捕捉图像的细节特征。其次，考虑到高频信息在图像细节恢复中的重要性，加入了频域损失函数，旨在频域上衡量重建图像与目标图像间的差异，从而有效地捕捉图像特征。最终，通过像素重排重建出高分辨率图像。实验表明，利用该算法提升绝缘子缺陷数据集图像质量后，在YOLO11算法中使得绝缘子检测精确率P提升1.6%，mAP提升2.1%。

（4）针对绝缘子检测精度低的问题，构建了改进YOLO11的检测算法。首先，在C2PSA模块中引入iRMB注意力，能够高效提取局部特征，捕捉全局上下文信息。其次，采用EM-Concat模块替换部分Concat操作，增强了特征图之间的联系。最后，使用基于改进的自适应空间特征融合ASFF实现多尺度检测，提升了绝缘子的检测精度。实验表明：改进算法相比YOLO11精确率P提升4.1%，mAP提升5.5%。最终，设计了输电线路绝缘子检测识别系统，能够准确地识别绝缘子缺陷，并将检测结果展示给用户，极大地提高了绝缘子的检测效率。

With the growing demand for ultra-high voltage and long-distance power transmission in China, the scale of overhead transmission line construction continues to expand. However, in the western regions, transmission lines are often built in regions with complex terrain such as the Loess Plateau. Insulator components of transmission lines are affected by strong winds, fog, and other environmental factors, leading to malfunctions such as damage and flashovers, which in turn impact transmission safety. At present, intelligent inspection of insulators mainly relies on perception devices combined with object detection algorithms. However, due to the impact of the complex environment and resource constraints, the quality of acquired insulator images fails to meet the requirements for intelligent detection, thus affecting the accuracy of insulator detection.To address this issue, this paper conducts research from two aspects: improving image quality and refining detection algorithms, with the aim of enhancing the accuracy of insulator detection. The main work is as follows：

(1) To address insulator defects like damage and flashover caused by strong winds and fog in the loess plateau terrain, an insulator defect dataset has been established and augmented using data augmentation techniques.

(2) To solve the problem of insulator image detail loss caused by fog, a feature - fusion - based cyclic GAN image - deraining algorithm is proposed. It incorporates FCANet frequency - channel attention into USID-Net, focusing the network on frequency features of rain information to enhance feature representation.In response to color shift in USID-Net derained images, color consistency loss is introduced. By measuring color - space similarity between derained and clear images, it effectively reduces color distortion.Lastly, DenseFuse is used to fuse deraining results from the improved USID-Net and AOD-Net, boosting deraining performance.Experiments show that compared to USID-Net and AOD-Net, the improved algorithm raises PSNR by 0.69 dB and 1.85dB.

(3) Due to restrictions in power supply, network transmission, etc., front-end perception devices lack sufficient resolution for insulator intelligent detection. So, a super-resolution reconstruction algorithm based on multi-scale residual network is proposed. A multi-scale feature attention module (MSFA) is designed, incorporating RCCA and FPA attention mechanisms. These mechanisms guide the model to accurately capture image detail features. Also, considering the significance of high-frequency information for image detail recovery, a frequency-domain loss function is introduced. It measures differences between reconstructed and target images in the frequency domain to effectively capture image features. Finally, high-resolution images are reconstructed via pixel rearrangement. Experiments show that this algorithm improves insulator detection accuracy. When applied to the YOLO11 algorithm for insulator defect detection, it boosts precision P by 1.6% and mAP by 2.1%.

(4) To tackle the issue of low insulator detection accuracy, an improved YOLO11 detection algorithm has been developed. Firstly, the iRMB attention mechanism is introduced into the C2PSA module. This allows for efficient extraction of local features and captures global context information. Secondly, the EM-Concat module is used to replace some Concat operations, enhancing the connections between feature maps. Finally, multi-scale detection is achieved through an improved Adaptive Spatial Feature Fusion (ASFF), which boosts the detection accuracy of insulators. Experiments have shown that compared to YOLO11, the improved algorithm increases precision P by 4.1% and mAP by 5.5%. A transmission line insulator detection and recognition system has also been designed. It can accurately identify insulator defects and present the detection results to users, thereby significantly improving the efficiency of insulator detection.

[1] 符春蕊. 输电线路杆塔基础施工技术和安全注意事项分析[J]. 人民黄河, 2024, 46(S1): 175-176.

[2] 夏祥泰, 曹枚根, 张若愚. 复合支柱绝缘子在长期载荷下的损伤识别方法[J]. 广东电力, 2024, 37(02): 95-102.

[3] 申迎松, 戴万波. 煤矿井下智能电力监控及故障诊断系统应用研究[J]. 工矿自动化, 2024, 50(S1): 200-204.

[4] 汤全, 杨倩. 西部黄土高原矿区采煤沉陷多源遥感监测技术进展与展望[J]. 煤炭科学技术, 2023, 51(12): 9-26.

[5] 张血琴,周志鹏,郭裕钧,等.不同材质绝缘子污秽等级高光谱检测方法研究[J]. 电工技术学报, 2023, 38(07): 1946-1955.

[6] 赵鹏,胡玉耀,蒋兴良,等.绝缘子不均匀积污成因及其对交流闪络特性的影响[J]. 电子测量与仪器学报, 2023, 37(12): 186-195.

[7] 吴雪茹.绝缘子缺陷智能检测算法及检测系统研究[D]. 安徽：安徽理工大学, 2024.

[8] 樊轶伦,陈蕾,张本科,等.基于多传感器的无人机配电网架空线路自主巡检和姿态控制[J]. 电测与仪表, 2024, 61(08): 186-194.

[9] 石玉东.以更高站位加快推进泛在电力物联网建设[J]. 国家电网, 2019, (09): 42-43.

[10] 孙启刚,宋卓彦,鉴庆之,等.基于无线传感网的输电铁塔结构健康状态监测技术及应用[J]. 工业建筑, 2022, 52(10): 224-228.

[11] 刘开培,李博强,秦亮,等.深度学习目标检测算法在架空输电线路绝缘子缺陷检测中的应用研究综述[J]. 高电压技术, 2023, 49(09): 3584-3595.

[12] 刘悦.复杂场景下无人机巡检输电线路绝缘子缺陷检测技术研究[D]. 西安: 西安理工大学, 2024.

[13] He K, Sun J, Tang X. Single image haze removal using dark channel prior[J]. IEEE transactions on pattern analysis and machine intelligence, 2010, 33(12): 2341-2353.

[14] Nishino K, Kratz L, Lombardi S. Bayesian defogging[J]. International journal of computer vision, 2012, 98: 263-278.

[15] Zhu Q, Mai J, Shao L. A fast single image haze removal algorithm using color attenuation prior[J]. IEEE transactions on image processing, 2015, 24(11): 3522-3533.

[16] Jobson D J, Rahman Z u, Woodell G A. Properties and performance of a center/surroundretinex[J]. IEEE Transactions on Image Processing, 1997, 6(3): 451-462.

[17] Cai B, Xu X, Jia K, et al. Dehazenet: An end-to-end system for single image haze removal[J]. IEEE transactions on image processing, 2016, 25(11): 5187-5198.

[18] Li B, Peng X, Wang Z, et al. AOD-Net: All-in-One Dehazing Network［C］// Proceedings of the IEEE international conference on computer vision. 2017: 4770-4778.

[19] Engin D, Genç A, Kemal Ekenel H. Cycle-Dehaze: Enhanced CycleGAN for Single Image Dehazing［C］// Proceedings of the IEEE conference on computer vision and pattern recognition workshops. 2018: 825-833.

[20] Zhao S, Zhang L, Shen Y, et al. RefineDNet: A weakly supervised refinement framework for single image dehazing[J]. IEEE Transactions on Image Processing, 2021, 30: 3391-3404.

[21] Li J, Li Y, Zhuo L, et al. USID-Net: Unsupervised single image dehazing network via disentangled representations[J]. IEEE transactions on multimedia, 2022, 25: 3587-3601.

[22] Li X, Yu H, Zhao C, et al. DADRNet: cross-domain image dehazing via domain adaptation anddisentangled representation[J]. Neurocomputing, 2023, 544: 126242.

[23] 闫在爽,贺鹏.基于物理解耦与自适应特征提取的无监督图像去雾[J/OL].现代电子技术, 2025-04-13.

[24] 苏腾华, 吕莉, 樊棠怀,等. 暗通道先验优化的生成对抗网络图像去雾算法[J]. 南昌工程学院学报, 2025, 44(01): 81-90.

[25] 张慧, 王新猛, 程占军, 等. 基于深度学习的雾天非法采砂船只辨识方法[J/OL]. 电子测量与仪器学报, 2025-04-13.

[26] Parker J A, Kenyon R V, Troxel D E. Comparison of interpolating methods for image resampling[J]. IEEE Transactions on medical imaging, 1983, 2(1): 31-39.

[27] Blu T, Thévenaz P, Unser M. Linear interpolation revitalized[J]. IEEE Transactions on Image Processing, 2004, 13(5): 710-719.

[28] Aly H A, Dubois E. Image up-sampling using total-variation regularization with a new observation model[J]. IEEE Transactions on Image Processing, 2005, 14(10): 1647-1659.

[29] Dong C, Loy C C, He K, et al. Image super-resolution using deep convolutional networks[J]. IEEE transactions on pattern analysis and machine intelligence, 2015, 38(2): 295-307.

[30] Shi W, Caballero J, Huszár F, et al. Real-time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network［C］// Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 1874-1883.

[31] Ledig C, Theis L, Huszár F, et al. Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network［C］// Proceedings of the IEEE conference on computer vision and pattern recognition. 2017: 4681-4690.

[32] 彭晏飞, 刘蓝兮, 王刚, 等. 基于拆分注意力网络的单图像超分辨率重建[J]. 液晶与显示, 2024, 39(07): 950-960.

[33] 邓酩, 柳庆龙, 侯立宪. 多尺度残差生成对抗网络的图像超分辨率重建[J]. 科学技术与工程, 2023, 23(31): 13472-13481.

[34] 盛月, 辛月兰, 王庆庆, 等. 基于注意力机制的多尺度融合图像超分辨率重建[J]. 激光杂志, 2024, 45(03): 118-125.

[35] 王进花, 魏婷, 曹洁, 等. 改进SwinIR的多特征融合图像超分辨率重建[J]. 西安电子科技大学学报, 2024, 51(06): 171-181.

[36] 程德强, 马祥, 寇旗旗, 等. 基于多路特征校准的轻量级图像超分辨率重建算法[J/OL].计算机辅助设计与图形学学报, 2025-04-13.

[37] Ding L, Rao Z Q, Ding B, et al. Research on defect detection method of railway transmission line insulators based on GC-YOLO[J]. IEEE Access, 2023, 11: 102635-102642.

[38] Souza B J, Stefenon S F, Singh G, et al. Hybrid-YOLO for classification of insulators defects in transmission lines based on UAV[J]. International Journal of Electrical Power & Energy Systems, 2023, 148: 108982.

[39] Yang Z, Xie R, Liu L, et al. Dense-YOLOv7: improved real-time insulator detection framework based on YOLOv7[J]. International Journal of Low-Carbon Technologies, 2024, 19: 157-170.

[40] 刘熹, 陈晨, 双丰. 基于改进YOLOv7-tiny的多种类绝缘子检测算法[J]. 仪器仪表学报, 2024, 45(09): 101-110.

[41] 白晓静, 谢雅祺, 赵淼, 等. 基于局部特征深度信息的绝缘子小样本缺陷检测[J]. 电网技术, 2024, 48(02): 740-752.

[42] 唐炳南, 柳明贤, 李华, 等. 基于改进YOLOv10的绝缘子缺陷检测算法[J/OL]. 南方电网技术, 2025-04-13.

[43] Liu L, Ke C, Lin H. Dark-Center Based Insulator Detection Method in Foggy Environment[J]. Applied Sciences, 2023, 13(12): 7264.

[44] Sun H, Huang L, Yu M, et al. Dual Constraint Parallel Multi-scale Attention Network for Insulator Detection in Foggy Scene[C]// Chinese Conference on Pattern Recognition and Computer Vision (PRCV). Singapore: Springer Nature Singapore, 2024: 287-300.

[45] 汤璐, 王淑青, 王年涛, 等. 基于改进YOLOX网络的雾天绝缘子缺陷检测[J]. 高压电器, 2024, 60(03): 223-228.

[46] 江灵雅. 电力巡检图像中绝缘子故障识别研究[D]. 安徽: 安徽理工大学, 2023.

[47] 丁丽丹, 胡君红, 胡聪. 基于绝缘子缺陷检测的模糊图像增强算法研究[J]. 激光杂志, 2022, 43(10): 56-61.

[48] Cong S, Zhou Y. A review of convolutional neural network architectures and their optimizations[J]. Artificial Intelligence Review, 2023, 56(3): 1905-1969.

[49] Li Z, Liu F, Yang W, et al. A survey of convolutional neural networks: analysis, applications, and prospects[J]. IEEE transactions on neural networks and learning systems, 2021, 33(12): 6999-7019.

[50] Goodfellow I, Pouget-Abadie J, Mirza M, et al. Generative adversarial networks[J]. Communications of the ACM, 2020, 63(11): 139-144.

[51] Porkodi S P, Sarada V, Maik V, et al. Generic image application using gans (generative adversarial networks): A review[J]. Evolving Systems, 2023, 14(5): 903-917.

[52] Dewi C, Chen R C, Liu Y T, et al. Synthetic Data generation using DCGAN for improved traffic sign recognition[J]. Neural Computing and Applications, 2022, 34(24): 21465-21480.

[53] Suganthy M, Krishnamoorthy R, Nagarajaiah K, et al. Lung cancer classification based on auxiliary classifier (WGAN) optimised with HOA from CT images[J]. IETE Journal of Research, 2024, 70(9): 7504-7512.

[54] Liu C, Gu J, Yao L, et al. Research on embroidery style migration model based on texture cycle GAN[J]. International Journal of Clothing Science and Technology, 2025, 37(1): 138-153.

[55] Hindarto D, Handayani E T E. Revolution in Image Data Collection: CycleGAN as a Dataset Generator[J]. Sinkron: jurnal dan penelitian teknik informatika, 2024, 8(1): 444-454.

[56] Lee Y H, Kim H J. Comparative Analysis of YOLO Series (from V1 to V11) and Their Application in Computer Vision[J]. Journal of the Semiconductor & Display Technology, 2024, 23(4): 190-198.

[57] Xuan Y, Zhang X Y, Li C, et al. LAM-YOLOv11 for UAV transmission line inspection: overcoming environmental challenges with enhanced detection efficiency[J]. Multimedia Systems, 2025, 31(2): 1-13.

[58] Qin Z, Zhang P, Wu F, et al. FcaNet: Frequency Channel Attention Networks［C］// Proceedings of the IEEE/CVF international conference on computer vision. 2021: 783-792.

[59] Hu J, Shen L, Sun G. Squeeze-and-excitation networks[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2018: 7132-7141.

[60] Li H, Wu X J. DenseFuse: A fusion approach to infrared and visible images[J]. IEEE Transactions on Image Processing, 2018, 28(5): 2614-2623.

[61] HAO S, ZHANG X, MA X, et al. Foreign object detection in coal mine conveyor belt based on CBAM-YOLOv5[J]. Journal of China Coal Society, 2022, 47(11): 4149-4158.

[62] Wang H, Wei Z, Tang Q, et al. Attention Guidance Distillation Network for Efficient Image Super-Resolution［C］// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2024: 6287-6296.

[63] Yang A, Wei Z, Wang J, et al. Multi-feature self-attention super-resolution network[J]. The Visual Computer, 2024, 40(5): 3473-3486.

[64] Chen J, Duanmu C, Long H. Large kernel frequency-enhanced network for efficient single image super-resolution[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2024: 6317-6326.

[65] Zhang J, Li X, Li J, et al. Rethinking Mobile Block for Efficient Attention-Based Models［C］// 2023 IEEE/CVF International Conference on Computer Vision (ICCV). IEEE Computer Society, 2023: 1389-1400.

[66] Jiang T, Liu L, Hu C, et al. An advanced method for surface damage detection of concrete structures in low-light environments based on image enhancement and object detection networks[J]. Advances in Bridge Engineering, 2024, 5(1): 1-17.