基于改进Libra-RCNN的输电线路绝缘子识别

doi:10.3969/j.issn.1008-0198.2022.02.008

湖南电力 ›› 2022, Vol. 42 ›› Issue (2): 44-49.doi: 10.3969/j.issn.1008-0198.2022.02.008

基于改进Libra-RCNN的输电线路绝缘子识别

闾海庆¹, 雷远华¹, 王静², 邢学敏³, 杨静⁴

1.中国能源建设集团湖南省电力设计院有限公司,湖南长沙 410007;
2.湖南省第三测绘院,湖南长沙 410004;
3.长沙理工大学交通运输工程学院,湖南长沙 410114;
4.中国水利水电第八工程局有限公司,湖南长沙 410004

收稿日期:2021-11-30 修回日期:2022-02-28 发布日期:2025-08-05
作者简介:闾海庆(1981),男,江苏人,硕士,高级工程师,主要研究方向为资源环境与遥感。
基金资助:
国家自然科学基金项目(42074033)

Transmission Line Insulator Identification Based on Improved Libra-RCNN

LYU Haiqing¹, LEI Yuanhua¹, WANG Jing², XING Xuemin³, YANG Jing⁴

1. China Energy Engineering Group Hunan Electric Power Design Institute Co., Ltd., Changsha 410007, China;
2. Hunan Third Surveying and Mapping Institute, Changsha 410004, China;
3. School of Traffic and Transportation Engineering, Changsha University of Science & Technology, Changsha 410114, China;
4. China Water Resources and Hydropower Eighth Engineering Bureau Co., Ltd., Changsha 410004, China

Received:2021-11-30 Revised:2022-02-28 Published:2025-08-05

摘要/Abstract

摘要： 针对无人机航拍输电线路识别绝缘子的定位精度和稳定性较差等问题,提出一种基于ASFF金字塔网络的Libra-RCNN绝缘子检测模型。首先,使用FRN归一化层替代原BN层,消除归一化层对训练批次大小依赖,增加模型学习效率;然后在Libra-RCNN算法金字塔中引入ASFF网络结构,有效解决特征金字塔内部不一致问题;最后借助GIoU交并比替代原IoU交并比,更好精确绝缘子位置。在Insulators_Datasets绝缘子数据集中,改进Libra-RCNN模型平均准确率达94.10%,召回率达97.51%;相较原Libra-RCNN模型分别提高2.23%、2.61%,表明所提算法能稳定、有效地识别绝缘子。

关键词: 绝缘子检测, Libra-RCNN模型, FRN归一化层, ASFF网络, GIoU交并比

Abstract: To solve the problems of poor positioning accuracy and stability of existing insulators identified by UAV aerial photography transmission lines, a Libra-RCNN insulator detection model based on ASFF pyramid network is proposed. Firstly, the FRN normalized layer is used to replace the original BN layer to eliminate the dependence of the normalized layer on the size of training batch and increase the learning efficiency of the model.Then, ASFF network structure is introduced into Libra-RCNN algorithm pyramid to effectively solve the problem of inconsistency inside the feature pyramid.Finally, GIoU crossover ratio is used to replace the original IoU crossover ratio to better accurate insulator position.In Insulators_Datasets data set, the average accuracy of improved Libra-RCNN model is 94.10%, and the recall rate is 97.51%.Compared with the original Libra-RCNN model, the improvement rate is 2.23% and 2.61% respectively, which indicates that the proposed algorithm can identify insulators stably and effectively.

Key words: insulator detection, Libra-RCNN model, filter response normalization(FRN) layer, adaptively spatial feature fusion(ASFF)network, generalized intersection over union

中图分类号:

TP391
TM854

闾海庆, 雷远华, 王静, 邢学敏, 杨静. 基于改进Libra-RCNN的输电线路绝缘子识别[J]. 湖南电力, 2022, 42(2): 44-49.

LYU Haiqing, LEI Yuanhua, WANG Jing, XING Xuemin, YANG Jing. Transmission Line Insulator Identification Based on Improved Libra-RCNN[J]. Hunan Electric Power, 2022, 42(2): 44-49.

参考文献

[1] 刘文清,易衍孜,陈小惠,等. 基于云平台的人脸识别在机房监控中的应用[J].湖南电力,2021,41(1): 52-55.
[2] 梁琼,汤奕雄,李弘运,等. 通信故障智能研判系统的设计与实现[J].湖南电力,2021,41(2): 46-48,64.
[3] 陈兴新,岳一石,程紫熠,等. 基于XGBoost算法的绝缘子污秽放电在线诊断方法研究[J].湖南电力,2021,41(2): 36-40.
[4] 唐明珠,赵琪,龙文,等. 基于梯度提升决策树的风电机组齿轮箱故障检测[J].湖南电力,2019,39(6): 52-56,60.
[5] 陈剑刚,姚璞,杨俊武,等. 无人机在架空输电线路巡检中的应用研究[J].湖南电力,2019,39(5): 74-77.
[6] 侯春萍,章衡光,张巍,等. 输电线路绝缘子自爆缺陷识别方法[J].电力系统及其自动化学报,2019,31(6): 1-6.
[7] REN S,HE K,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.
[8] 蒋姗,孙渊,严道森.基于深度学习算法的航拍巡检图像的绝缘子识别[J].福州大学学报(自然科学版),2021,49(1): 58-64.
[9] Navaneeth Bodla,Bharat Singh, Rama Chellappa,et al.Soft-NMS Improving Object Detection With One Line of Code[EB/OL].(2017-04-05)[2021-11-28].https://arxiv.org/pdf/1704.04503v2.pdf.
[10] 周自强,赵淳,范鹏.基于多尺度特征融合Faster R-CNN的绝缘子自爆缺陷研究[J].水电能源科学,2020,38(11): 187-189,44.
[11] 易继禹,陈慈发,龚国强.基于改进Faster RCNN的输电线路航拍绝缘子检测[J].计算机工程.2021,47(6): 292-208,304.
[12] LIU Songtao,DI HUANG,WANG Yunhong. Learning Spatial Fusion for Single-Shot Object Detection[EB/OL].(2019-11-21)[2021-11-28]. https://arxiv.org/pdf/1911.09516.pdf.
[13] PANG J, CHEN K, SHI J, et al. Libra R-CNN: Towards Balanced Learning for Object Detection[EB/OL].(2019-06-20)[2021-11-28].https://arxiv.org/pdf/1904.02701.pdf.
[14] SAURABH Singh, SHANKAR Krishnan.Filter Response Normalization Layer: Eliminating Batch Dependence in the Training of Deep Neural Networks[EB/OL].(2019-11-09)[2021-11-28].https://arxiv.org/pdf/1911.09737.pdf.
[15] IOFFE S, SZEGEDY C. Batch normalization: Accelerating deep network training by reducing internal covariate shift[EB/OL].(2015-02-13)[2021-11-28].https://arxiv.org/pdf/1502.03167.pdf.
[16] HAMID Rezatofighi, NATHAN Tsoi, JunYoung Gwak,et al. Generalized Intersection over Union: A Metric and A Loss for Bounding Box Regression[EB/OL].(2019-11-19)[2021-11-28].https://arxiv.org/pdf/1911.08287.pdf.
[17] WUYuxin,HE Kaiming. Group Normalization[EB/OL].(2018-02-24)[2021-11-28]. https://arxiv.org/pdf/1803.08494v3.pdf.
[18] HU Jie , LI Shen, GABG Sun.Squeeze-and-Excitation Networks[EB/OL].(2018-10-25)[2021-11-28].https://arxiv.org/pdf/1709.01507v4.pdf.
[19] REDMON J, FARHAD A. YOLO9000: better, faster, stronger[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).2017,6517-6525.
[20] REDMON J,FARHADI A. YOLOv3: an incremental improvement[EB/OL].(2018-04-23)[2021-11-28].https://arxiv.org/pdf/1804.02767.pdf.

基于改进Libra-RCNN的输电线路绝缘子识别

Transmission Line Insulator Identification Based on Improved Libra-RCNN

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 4

编辑推荐

Metrics

本文评价

[1]	瞿旭, 龙崦平, 张伟, 袁超雄. 基于位图和矢量图联合识别的电气二次图纸解析方法研究[J]. 湖南电力, 2024, 44(6): 97-103.
[2]	高攀, 李飞, 彭远豪, 张璨辉, 彭海君. 基于jieba中文分词的电力客户精准分类方法[J]. 湖南电力, 2023, 43(5): 151-154.
[3]	李容嵩, 彭凌烟, 蒋成, 张定军. 基于区块链和属性加密的工程检测数据安全共享技术研究[J]. 湖南电力, 2022, 42(3): 65-72.
[4]	王生祺, 陈海波, 叶金翔. 基于深度学习与局部感知注意力的避雷器套管红外图像分割方法[J]. 湖南电力, 2025, 45(2): 122-128.