A Pantograph Horn Detection Method Based on Deep Learning Network

A good contact between the pantograph and catenary ensures the safety of high-speed train operation. Pantograph horn, which is the curved structure at both ends of the pantograph, plays important roles in monitoring the operation state of the train. Nowadays, deep learning method has a significant effect in the detection of horns and fault. In this paper, a pantograph horn detection method has been proposed. The method is based on single-shot mutibox detector(SSD) method, which is a real-time method and also with high detection accuracy. A on-orbit image data set with multiple viewing angles and multiple pantograph types is collected to be used in the training stage. The target region is converged through the combination of the feature map in early convolution layers and the prior knowledge. Then, detection results with the partial image and global image as input are obtained, and high accuracy detecting result is generated after confidential decision. Results on actual datasets show that our method can stably obtain accurate horn location, and help to monitor the pantograph status. Moreover, pantograph defects of several common pantograph types can be detected robustly.

[1]  Yi Li,et al.  R-FCN: Object Detection via Region-based Fully Convolutional Networks , 2016, NIPS.

[2]  Ross B. Girshick,et al.  Fast R-CNN , 2015, 1504.08083.

[3]  Zhigang Liu,et al.  Thermal Distribution Modeling and Experimental Verification of Contact Wire Considering the Lifting or Dropping Pantograph in Electrified Railway , 2016, IEEE Transactions on Transportation Electrification.

[4]  Ali Farhadi,et al.  You Only Look Once: Unified, Real-Time Object Detection , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[5]  Yang Song,et al.  Detection of Contact Wire Irregularities Using a Quadratic Time–Frequency Representation of the Pantograph–Catenary Contact Force , 2016, IEEE Transactions on Instrumentation and Measurement.

[6]  Trevor Darrell,et al.  Caffe: Convolutional Architecture for Fast Feature Embedding , 2014, ACM Multimedia.

[7]  Xiang Zhang,et al.  OverFeat: Integrated Recognition, Localization and Detection using Convolutional Networks , 2013, ICLR.

[8]  Hironobu Fukai,et al.  Automatic correction of measurement position using robust matching in contact wire inspection system , 2016, 2016 11th France-Japan & 9th Europe-Asia Congress on Mechatronics (MECATRONICS) /17th International Conference on Research and Education in Mechatronics (REM).

[9]  Trevor Darrell,et al.  Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation , 2013, 2014 IEEE Conference on Computer Vision and Pattern Recognition.

[10]  Mehmet Karaköse,et al.  A New Experimental Approach Using Image Processing-Based Tracking for an Efficient Fault Diagnosis in Pantograph–Catenary Systems , 2017, IEEE Transactions on Industrial Informatics.

[11]  Mehmet Karaköse,et al.  Anomaly detection using a modified kernel-based tracking in the pantograph-catenary system , 2015, Expert Syst. Appl..

[12]  Wei Liu,et al.  SSD: Single Shot MultiBox Detector , 2015, ECCV.

[13]  Yang Song,et al.  Ensemble EMD-Based Automatic Extraction of the Catenary Structure Wavelength From the Pantograph–Catenary Contact Force , 2016, IEEE Transactions on Instrumentation and Measurement.

[14]  Mehmet Karaköse,et al.  A new arc detection method based on fuzzy logic using S-transform for pantograph–catenary systems , 2018, J. Intell. Manuf..