|
|
|
| Insulator defect detection method based on Transformer and information fusion |
| CHEN Tianhang1, ZENG Yezhan1, DENG Qian2, ZHONG Chunliang1 |
1. College of Electrical and Information Engineering, Hu’nan University of Technology, Zhuzhou, Hu’nan 412007;;
2. College of Rail Transit, Hu’nan University of Technology, Zhuzhou, Hu’nan 412007 |
|
|
|
|
Abstract Aiming at the existing insulator aerial images, which have complex backgrounds and are difficult to detect flashover and broken defects, a global and local information fusion (GLIF)-you only look once v8s (YOLOv8s) insulator detection algorithm is proposed. The algorithm uses EfficientFormerV2 as the backbone network to improve the model’s ability to extract global information. A feature enhancement module (FEM) is designed based on global and local information to reduce the loss of deep network information through information fusion. Ablation experiments and comparison experiments are carried out on insulators defects dataset, and the experimental results show that the proposed algorithm achieves 77.5% class-wide average accuracy, and its flashover and broken defect detection accuracy reaches 67.7% and 73.5%. Compared with other mainstream algorithms, the detection frame confidence of the proposed algorithm is higher.
|
|
Received: 25 April 2024
|
|
|
|
| Cite this article: |
|
CHEN Tianhang,ZENG Yezhan,DENG Qian等. Insulator defect detection method based on Transformer and information fusion[J]. Electrical Engineering, 2024, 25(8): 11-17.
|
|
|
|
| URL: |
|
http://dqjs.cesmedia.cn/EN/Y2024/V25/I8/11
|
[1] 安灵旭, 唐其筠, 李中成, 等. 人工智能在配电网运维中的应用研究[J]. 电气技术, 2019, 20(10): 103-106.
[2] 梁新福, 罗日成, 党世轩, 等. 基于数字图像处理的电力线异物识别方法研究[J]. 电气技术, 2022, 23(2): 73-78.
[3] 徐晓宇. 高压输电线路巡检图像缺陷检测算法研究[D]. 杭州: 杭州电子科技大学, 2019.
[4] 李岩. 基于HOG特征和SVM的绝缘子识别与定位[J]. 交通运输工程与信息学报, 2015, 13(4): 53-60.
[5] 董懿飞, 舒胜文, 陈诚, 等. 基于红外图像分割与SSA-SVM的复合绝缘子缺陷检测方法[J]. 电气技术, 2021, 22(11): 73-79.
[6] 臧国强, 刘晓莉, 徐颖菲, 等. 深度学习在电力设备缺陷识别中的应用进展[J]. 电气技术, 2022, 23(6): 1-7.
[7] 何文其, 李剑, 赵文浩. 基于改进FasterRCNN的绝缘子异常检测[J]. 浙江电力, 2021, 40(8): 40-46.
[8] 李斌, 屈璐瑶, 朱新山, 等. 基于多尺度特征融合的绝缘子缺陷检测[J]. 电工技术学报, 2023, 38(1): 60-70.
[9] 周宸, 高伟, 郭谋发. 基于YOLOv4模型的玻璃绝缘子自爆缺陷识别方法[J]. 电气技术, 2021, 22(5): 38-42, 49.
[10] 杨露露, 马萍, 王聪, 等. 结合特征重用与特征重建的YOLO绝缘子检测方法[J]. 计算机工程, 2024, 50(7): 303-313.
[11] 田永林, 王雨桐, 王建功, 等. 视觉Transformer研究的关键问题: 现状及展望[J]. 自动化学报, 2022, 48(4): 957-979.
[12] 苟军年, 杜愫愫, 刘力. 基于改进掩膜区域卷积神经网络的输电线路绝缘子自爆检测[J]. 电工技术学报, 2023, 38(1): 47-59.
[13] LI Yanyu, HU Ju, WEN Yang, et al.Rethinking vision transformers for MobileNet size and speed[C]//2023 IEEE/CVF International Conference on Computer Vision (ICCV), Paris, France, 2023: 16843-16854.
[14] LIN T Y, DOLLAR P, GIRSHICK R, et al.Feature pyramid networks for object detection[C]//2017 IEEE Conference on Computer Vision and Pattern Recog- nition (CVPR), Honolulu, HI, USA, 2017: 936-944.
[15] LIU Shu, QI Lu, QIN Haifang, et al.Path aggregation network for instance segmentation[C]//2018 IEEE/ CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, UT, USA, 2018: 8759-8768.
[16] NARAYANAN M. SENetV2: aggregated dense layer for channelwise and global representations[EB/OL]. 2023: arXiv preprint arXiv: 2311.10807. http://arxiv. org/abs/2311.10807.
[17] WANG C Y, YEH I H, LIAO H Y M. YOLOv9: learning what you want to learn using pogrammable gradient information[EB/OL]. 2024: arXiv preprint arXiv: 2402.13616. http://arxiv.org/abs/2402.13616. |
|
|
|
2024, Vol. 25 
