多尺度融合与动态自校正旋转的吊弦检测算法

doi:10.6040/j.issn.1672-3961.0.2025.093

摘要/Abstract

摘要： 针对高速铁路接触网中因吊弦松弛、断裂严重影响列车正常运行的问题,提出一种基于YOLOv8n的多尺度动态旋转(multi-scale dynamic rotation YOLOv8n, MDR-YOLOv8n)算法,用于检测吊弦的异常状态。通过高速铁路接触网4C检测系统获取高清吊弦图像,进行图像扩充;设计一种卷积局部注意力机制(convolutional local attention version 2, CloAttV2)并嵌入跨阶段部分融合(cross stage partial fusion, C2f)主干网络,通过轴向自适应池化与动态稀疏注意力门控协同作用,强化全局与局部特征融合,增强对吊弦关键特征的捕捉能力;设计一种含自校正机制的多尺度特征融合轻量化动态上采样模块,通过自适应调整特征图的采样权重,有效利用上下文语义信息,降低模型参数量,显著提升抗干扰能力;设计面向旋转框的任务对齐动态检测头(oriented bounding box-task align dynamic detection head, OBB-TADDH),采用任务对齐机制优化旋转目标定位效果,减少冗余信息,提高小目标检测能力。试验结果表明,MDR-YOLOv8n在置信度0.5下的平均精度较YOLOv8n模型提升3.7百分点,推理速度提升2.3百分点,在复杂环境下能保持较高的检测性能。MDR-YOLOv8n在检测精度、推理速度和轻量化方面能够优化平衡关系,为4C检测系统的智能升级提供新方案。

关键词: YOLOv8n, 接触网吊弦, 动态上采样, 注意力机制, 旋转目标检测

Abstract: To address the issue of catenary dropper slackness and fractures in high-speed railway contact networks, which severely disrupt train operations, a multi-scale dynamic rotation YOLOv8n(MDR-YOLOv8n)algorithm was proposed to detect the abnormal states of the catenary dropper. High-resolution dropper images were acquired through the high-speed railway contact network 4C inspection system and enhanced via data augmentation. A convolutional local attention version 2(CloAttV2)was designed and integrated into the cross stage partial fusion(C2f)backbone network. Through collaborative axial adaptive pooling and dynamic sparse attention gating, the effectiveness of global-local feature fusion was boosted while the capture capability of key dropper features was enhanced. A lightweight multi-scale dynamic upsampling module with self-calibration mechanism was designed to adaptively adjust sampling weights of the feature maps, effectively utilizing contextual semantic information while reducing model parameters and enhancing anti-interference capability. An oriented bounding box-task align dynamic detection head(OBB-TADDH)was designed, which leveraged task-aligned optimization to enhance rotated object localization, suppress feature redundancy, and improve detection sensitivity for small targets. Experimental results demonstrated that MDR-YOLOv8n achieved 3.7 percentage points improvement in mean average precision at a confidence threshold of 0.5 and 2.3 percentage points increase in inference speed compared to the YOLOv8n model, while maintaining high detection performance under complex environmental conditions. MDR-YOLOv8n optimized the balance among detection accuracy, inference speed, and lightweight design, providing a novel solution for the intelligent upgrade of 4C inspection system.

Key words: YOLOv8n, overhead contact system dropper, dynamic upsample, attention mechanism, rotating target detection

中图分类号:

TP391

赵峰,刘瑞,王英,陈小强,葛磊蛟,马爱平. 多尺度融合与动态自校正旋转的吊弦检测算法[J]. 山东大学学报 (工学版), 2026, 56(2): 1-10.

ZHAO Feng, LIU Rui,WANG Ying, CHEN Xiaoqiang, GE Leijiao, MA Aiping. Multi-scale fusion and dynamic self-calibrating rotation-based catenary dropper detection algorithm[J]. Journal of Shandong University(Engineering Science), 2026, 56(2): 1-10.

参考文献

[1] YANG Y, GENG S P, CHENG C, et al. An edge algorithm for assessing the severity of insulator discharges using a lightweight improved YOLOv8[J]. Journal of Electrical Engineering & Technology, 2025, 20(1): 807-816.
[2] ZHOU F, HE F J, GUI C C, et al. SAR target detection based on improved SSD with saliency map and residual network[J]. Remote Sensing, 2022, 14(1): 180.
[3] MA H Y, YANG B H, WANG R R, et al. Automatic extraction of discolored tree crowns based on an improved Faster-RCNN algorithm[J]. Forests, 2025, 16(3): 382.
[4] DEWANGAN S K, CHOUBEY S, PATRA J, et al. IMU-CNN: implementing remote sensing image restoration framework based on mask-upgraded cascade R-CNN and deep autoencoder[J]. Multimedia Tools and Applications, 2024, 83(27): 69049-69081.
[5] 齐冬莲, 钱佳莹, 闫云凤, 等. 一种基于RefineDet网络和霍夫变换的高速铁路接触网吊弦状态多尺度检测方法[J]. 电子与信息学报, 2021, 43(7): 2014-2022. QI Donglian, QIAN Jiaying, YAN Yunfeng, et al. A multi-scale detection method for dropper states in high-speed railway contact network based on RefineDet network and Hough transform[J]. Journal of Electronics & Information Technology, 2021, 43(7): 2014-2022.
[6] 李雪峰, 刘海莹, 刘高华, 等. 基于深度学习的输电线路销钉缺陷检测[J]. 电网技术, 2021, 45(8): 2988-2995. LI Xuefeng, LIU Haiying, LIU Gaohua, et al. Transmission line pin defect detection based on deep learning[J]. Power System Technology, 2021, 45(8): 2988-2995.
[7] 顾桂梅, 贾耀华, 温柏康. 基于YOLOv5s的接触网吊弦线和载流环缺陷识别算法[J]. 铁道科学与工程学报, 2023, 20(3): 1066-1076. GU Guimei, JIA Yaohua, WEN Bokang. Defect identification algorithm of dropper line and current-carrying ring of catenary based on YOLOv5s[J]. Journal of Railway Science and Engineering, 2023, 20(3): 1066-1076.
[8] 卞建鹏, 薛秀茹, 崔跃华, 等. 基于EfficientDet与Vision Transformer的接触网吊弦故障检测[J]. 铁道科学与工程学报, 2023, 20(6): 2340-2349. BIAN Jianpeng, XUE Xiuru, CUI Yuehua, et al. Fault detection of catenary hanger based on EfficientDet and Vision Transformer[J]. Journal of Railway Science and Engineering, 2023, 20(6): 2340-2349.
[9] 王晓明, 陈智宇, 董文涛, 等. 基于YOLOv7x的接触网吊弦缺陷检测方法[J]. 华东交通大学学报, 2024, 41(3): 65-73. WANG Xiaoming, CHEN Zhiyu, DONG Wentao, et al. Detection method of catenary hanging string based on YOLOv7x[J]. Journal of East China Jiaotong University, 2024, 41(3): 65-73.
[10] 李瑞生, 张彦龙, 翟登辉, 等. 基于改进SSD的输电线路销钉缺陷检测[J]. 高电压技术, 2021, 47(11): 3795-3802. LI Ruisheng, ZHANG Yanlong, ZHAI Denghui, et al. Pin defect detection of transmission line based on improved SSD[J]. High Voltage Engineering, 2021, 47(11): 3795-3802.
[11] XIE G B, XU Z J, LIN Z Y, et al. GRFS-YOLOv8: an efficient traffic sign detection algorithm based on multiscale features and enhanced path aggregation[J]. Signal, Image and Video Processing, 2024, 18(6/7): 5519-5534.
[12] XIE T Y, JIANG Y T, WANG C L, et al. A drone system for PV panel cleaning based on YOLOv8 and automated flight path planning[J]. World Scientific Research Journal, 2025, 11(2): 47-54.
[13] 董明书, 陈俐企, 马川义, 等. 沥青路面内部裂缝雷达图像智能判识算法研究[J]. 山东大学学报(工学版), 2025, 55(3): 72-79. DONG Mingshu, CHEN Liqi, MA Chuanyi, et al. Deep learning-based intelligent judgment for radar detection of pavement cracks[J]. Journal of Shandong University(Engineering Science), 2025, 55(3): 72-79.
[14] 袁博雅, 李尧, 叶青. 面向输电线路绝缘子的GER-YOLO缺陷检测算法[J]. 激光与光电子学进展, 2024, 61(22): 2212005. YUAN Boya, LI Yao, YE Qing.GER-YOLO fault-detection algorithm for transmission-line insulators[J]. Laser & Optoelectronics Progress, 2024, 61(22): 2212005.
[15] 栗莎, 王永雄, 王哲, 等. 融合局部和全局特征的铸件缺陷检测[J]. 应用科学学报, 2024, 42(5): 757-768. LI Sha, WANG Yongxiong, WANG Zhe, et al. Casting defect detection based on local and global features[J]. Journal of Applied Sciences, 2024, 42(5): 757-768.
[16] WANG T, CHEN Q M, LANG X, et al. Detection of oscillations in process control loops from visual image space using deep convolutional networks[J]. IEEE/CAA Journal of Automatica Sinica, 2024, 11(4): 982-995.
[17] KIM W J, KIM I, LEE S H. Enhancing the performance of the neural network model for the EMG regression case using Hadamard product[J]. Journal of Mechanical Science and Technology, 2024, 38(7): 3607-3613.
[18] 李淇, 石艳, 范桃. 改进YOLOv8n的O型密封圈表面缺陷检测算法研究[J]. 计算机工程与应用, 2024, 60(18): 126-135. LI Qi, SHI Yan, FAN Tao. Research on O-ring surface defect detection algorithm based on improved YOLOv8n[J]. Computer Engineering and Applications,2024, 60(18): 126-135.
[19] 曲优, 李文辉. 基于锚框变换的单阶段旋转目标检测方法[J]. 吉林大学学报(工学版), 2022, 52(1): 162-173. QU You, LI Wenhui. Single-stage rotated object detection network based on anchor transformation[J]. Journal of Jilin University(Engineering and Technology Edition), 2022, 52(1): 162-173.
[20] 王亚彬, 徐爱俊, 周素茵, 等. 基于Byte的生猪多目标跟踪算法[J]. 农业工程学报, 2025, 41(7): 145-155. WANG Yabin, XU Aijun, ZHOU Suyin, et al. Multi-object tracking of pig behavior using Byte algorithm[J]. Transactions of the Chinese Society of Agricultural Engineering, 2025, 41(7): 145-155.
[21] 朱玉敏, 孙光灵, 缪飞. 基于改进YOLOv8算法的鱼眼图像下行人检测[J]. 计算机科学与探索, 2025, 19(2): 443-453. ZHU Yumin, SUN Guangling, MIAO Fei. Pedestrian detection in fisheye images based on improved YOLOv8 algorithm[J]. Journal of Frontiers of Computer Science and Technology, 2025, 19(2): 443-453.
[22] 杨茜, 熊炜, 孟圣哲, 等. 基于改进YOLOv8的绝缘子缺陷检测方法[J]. 电子测量技术, 2025, 48(7): 86-97. YANG Qian, XIONG Wei, MENG Shengzhe, et al. Insulator defect detection method based on improved YOLOv8[J]. Electronic Measurement Technology, 2025, 48(7): 86-97.
[23] XU S H, WANG J H, HE N, et al. Optimizing under water image enhancement: integrating semi-supervised learning and multi-scale aggregated attention[J]. The Visual Computer, 2025, 41(5): 3437-3455.
[24] SU Q H, MU J H. Complex scene occluded object detection with fusion of mixed local channel attention and multi-detection layer anchor-free optimization[J]. Automation, 2024, 5(2): 176-189.
[25] 郑云水, 蒙阳. 基于改进YOLOv8s的铁路车站信号平面布置图信息提取方法[J]. 中国铁道科学, 2024, 45(5): 209-220. ZHENG Yunshui, MENG Yang. Information extraction method of railway station signal plan layout based on improved YOLOv8s[J]. China Railway Science, 2024, 45(5): 209-220.
[26] 肖振久, 严肃, 曲海成. 基于多重机制优化YOLOv8的复杂环境下安全帽检测方法[J]. 计算机工程与应用, 2024, 60(21): 172-182. XIAO Zhenjiu, YAN Su, QU Haicheng. Safety helmet detection method in complex environment based on multi-mechanism optimization of YOLOv8[J]. Computer Engineering and Applications, 2024, 60(21): 172-182.

多维度评价

Viewed

Full text

Abstract

Cited

Shared

Discussed