MAID: making accurate transmission line icing detector by enhancing inaccurate dataset

Abstract

Power transmission lines are a critical component of the entire power system, and ice accretion incidents caused by various types of power systems can result in immeasurable harm. Currently, network models used for ice detection on power transmission lines require a substantial amount of sample data to support their training, and their drawback is that detection accuracy is significantly affected by the inaccurate annotation among training dataset. Therefore, we propose a transformer-based detection model, structured into two stages to collectively address the impact of inaccurate datasets on model training. In the first stage, a spatial similarity enhancement (SSE) module is designed to leverage spatial information to enhance the construction of the detection framework, thereby improving the accuracy of the detector. In the second stage, a target similarity enhancement (TSE) module is introduced to enhance object-related features, reducing the impact of inaccurate data on model training, thereby expanding global correlation. Additionally, by incorporating a multi-head adaptive attention window (MAAW), spatial information is combined with category information to achieve information interaction. Simultaneously, a quasi-wavelet structure, compatible with deep learning, is employed to highlight subtle features at different scales. Experimental results indicate that the proposed model in this paper outperforms existing mainstream detection models, demonstrating superior performance and stability.

Keywords

A

References

[1] ZhangY, QiuC, YangF, et al.. Overview of the application of deep learning in power grid image data and space-time data. Power system technology, 2019, 43(6): 1865-1873[J]

[2] HaoY, HuangL, WeiJ, et al.. The detecting system and method of quasi-distributed fiber Bragg grating for overhead transmission line conductor ice and composite insulator icing load. IEEE transactions on power delivery, 2022, 38(3): 1799-1809. J]

[3] LiL, LuoD, YaoW. Analysis of transmission line icing prediction based on CNN and data mining technology. Soft computing, 2022, 26(16): 7865-7870. J]

[4] LiB, BaiJ, HeJ, et al.. A review on superhydrophobic surface with anti-icing properties in overhead transmission lines. Coatings, 2023, 132301. J]

[5] XieB, ZhangC, GongQ W, et al.. Icing thickness prediction of overhead power transmission lines using parallel coordinates and convolutional neural networks. Advances in Computer Science for Engineering and Education III: Proceedings of the International Conference on Computer Science for Engineering and Education (ICCSEE 2019), March 29–31, 2019, Zhengzhou, China, 2019, Cham. Springer International Publishing. 255267[C]

[6] YangL, ChenJ, HaoY, et al.. Experimental study on ultrasonic detection method of ice thickness for 10 kV overhead transmission lines. IEEE transactions on instrumentation and measurement, 2023, 72: 1-10[J]

[7] HeK, GkioxariG, DollárP, et al.. Mask R-CNN. Proceedings of the IEEE International Conference on Computer Vision (ICCV), October 22–November 2, 2017, Venice, Italy, 2017, New York. IEEE. 29612969[C]

[8] ZhangQ, XuY, ZhangJ, et al.. VSA: learning varied-size window attention in vision transformers. European Conference on Computer Vision (ECCV 2022), October 23–27, 2022, Tel Aviv, Israel, 2022, Cham. Springer Nature Switzerland. 466483[C]

[9] AgarwalA, AroraC. Attention everywhere: monocular depth prediction with skip attention. Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV 2023), January 2–7, 2023, Waikoloa, HI, USA, 2023, New York. IEEE. 58615870[C]

[10] LI P, ZHAO N, ZHOU D, et al. Multivariable time series prediction for the icing process on overhead power transmission line[J]. The scientific world journal, 2014.

[11] LiuC, WangK, LuH, et al.. Robust object detection with inaccurate bounding boxes. European Conference on Computer Vision (ECCV 2022), October 23–27, 2022, Tel Aviv, Israel, 2022, Cham. Springer Nature Switzerland. 5369[C]

[12] ZhangX, YangY, FengJ. Learning to localize objects with noisy labeled instances. Proceedings of the AAAI Conference on Artificial Intelligence (AAAI-19), January 27–February 1, 2019, Honolulu, Hawaii, USA, 2019, Menlo Park. AAAI Press. 92199226[C]

[13] RenS, HeK, GirshickR, et al.. Faster R-CNN: towards real-time object detection with region proposal networks. Advances in Neural Information Processing Systems 28 (NeurIPS 2015), December 7–12, 2015, Montreal, Canada, 2016, New York. Curran Associates Inc.. 11371149[C]

[14] ZHU X, SU W, LU L, et al. Deformable DETR: deformable transformers for end-to-end object detection[EB/OL]. (2020-10-08) [2023-12-05]. http://arxiv.org/abs/2010.04159.

RIGHTS & PERMISSIONS

Tianjin University of Technology