Fault identification of catenary dropper based on improved CapsNet
Fault identification of catenary dropper based on improved CapsNet
Traditional fault identification algorithms applied to catenary dropper suffer from various problems due to itssmall contact area. These problems include misidentification and lower recognition rate of the faulty dropper. Comparedwith the traditional convolutional neural network, the vector is utilized as the input of the capsule network (CapsNet)for the first time, which can well retain the feature information such as the direction and angle of the target, and ismore suitable for identifying the dropper under complex background. Therefore, this paper proposes a dropper faultidentification algorithm based on improved capsule network. The convolutional layer of traditional 9×9 capsule networkis simplified through 1 × 1 reduction layer and 3 × 3 convolutional layer, and the optimization algorithm is adopted forparameter optimization to shorten the training weight time. At the same time, the output can retain more informationsuch as direction and angle, which can accurately identify the breakage and falling of current carrying broken. Thus,in order to better improve the accuracy and real-time of detecting the fault dropper from a running train operation, adropper fault identification algorithm based on an improved CapsNet is proposed in this paper. Experimental resultsshow that the improved CapsNet is well-suited for fault identification of catenary dropper, as it can effectively removethe interference caused by the complex background on the dropper image, and identify the image containing the faultydropper with a higher recognition rate.
___
- [1] Cho SH, Kim KH, Hong S. Efective object identication and association by varying coverage through rfid power
control. Journal of Computer Science & Technology 2014; 29 (1): 4-20.
- [2] Liu M, Cheng L, Huang D. A high reliability power supply path search method based on operation reliability theory.
In: Conference of Transactions of China Electrotechnical Society; Beijing, China; 2019. pp. 3004-3011.
- [3] Ou ZX. Research on cases of contact network electrical flashover and fault detection database. Measurement and
Control Technology 2018; 20(3): 107-110.
- [4] Yang F, Sun T, Song S. Research on key technology of automatic pharmaceutical container inspection machine
based on machine vision. Chinese Journal of Scientific Instrument 2008; 6(3): 562-566.
- [5] Yan W, Wang W, Su Q. Research on intelligent inspection technology of UAV based on 3D Information of
Transmission Lines. Chinese High-tech 2018; 19(01): 76-78.
- [6] Wang Y, Zhu LQ, Yu Zj. Scene segmentation and recognition algorithm for high-speed railway. Optics Journal 2019;
3(4): 1-14.
- [7] Nasi R, Honkavaara E, Tuominen S, Saari H, Pölönen I et al. Based on tree species identification using the novel
FPI based hyperspectral Cameras in Visible, NIR and SWIR spectral ranges. In: Conference Proceedings of Remote
Sensing and Spatial Information Sciences; Prague, Czech Republic; 2016. pp. 1143-1148.
- [8] Tian Y, Zhou QM. A robust non-uniform gray-scale image segmentation algorithm. Journal of Electronics and
Information Technology 2011; 33(06): 12-17.
- [9] Lin G, Wang B, Peng H. Multi-target detection and location of power line inspection image based on improved
Faster-RCNN. Electrical Automation Equipment 2018; 51(9): 21-27.
- [10] Shize H, Yachan Z, Miaomiao Z, Xiaoxue H. Arc detection and recognition in pantograph–catenary system based
on convolutional neural network. Information Sciences 2019; 501(3): 1024-1029.
- [11] Hinton GE, Osindero S, Teh YW. A fast learning algorithm for deep belief nets. Neural Computation 2006; 18(7):
1527-1554.
- [12] Yang CJ, Wang L. Identification of targets beyond the atmosphere based on probabilistic neural network. Chinese
Optics Letters 2011; 9(7): 1-5.
- [13] Lee SJ, Ban J, Choi H et al. Localization of slab identification numbers using deep learning. In: IEEE 2016
International Conference on Control, Automation and Systems (ICCAS); Xian, China; 2016. pp.1350-1369.
- [14] Li M, An Y. Ground moving target identification based on neural network. IEEE International Conference on
Internet Computing & Information Services 2011; 23(3): 15-26.
- [15] Gao JY, Yang XC. Robust visual tracking method based on deep learning. Journal of Computer Science 2017; 2(7):
1419-1434.
- [16] Li N. Pavement crack identification research based on deep learning framework Caffe. Engineering Technology
Research 2017; 17(2): 66-73.
- [17] Sabour S, Frosst N, Hinton G E. Dynamic routing between capsules. Computer Vision and Pattern Recognition
2017; 3(1): 90-96.
- [18] Minjong K, Suyoung C. Detection of centerline crossing in abnormal driving using CapsNet. Journal of Super
Computing 2018; 1(3) : 35-40.
- [19] Yuan ying. Fingerprint identification of capsule neural network based on MATLAB. Software 2019; 40(10): 89-92.
- [20] Jun J, Chen X. Improved sequential overlapping handwritten digit recognition method for capsule network. Laser
Magazine 2019; 40(07): 43-46.