Application of point cloud data in geometric detection of critical nodes of turnouts

doi:10.13474/j.cnki.11-2246.2024.0318

Abstract

Abstract: The process of detecting the operational status of a turnout is complex, and traditional methods require the use of rail inspection vehicles, rail gauges, support gauges, and height gauges for measurement. The equipment types are varied, the measurement process is lengthy, and there is a high demand for the time window. To improve detection efficiency, this paper proposes a key node measurement method for single turnout based on point clouds. The method utilizes CAD graphic elements and graph convolutional neural networks to achieve the precise, automated identification, segmentation, and extraction of the three-dimensional point cloud data of the turnout structure, with an accuracy rate of 99.68%. At the same time, by combining the geometric prior information of the turnout structure, the key geometric position parameters such as the track gauge, curve radius, support spacing, and rail elevation drop are accurately extracted in a rapid and precise manner. Verified by examples, the measurement results of the key geometric position detection method for turnouts based on point cloud proposed in this paper have a sub-millimeter error with the real value, which meets the requirements of actual engineering inspection, eliminates a variety of inspection equipment and saves a lot of time for skylight, which has a high degree of practicability, and it is the development trend of future turnout measurements.

Key words: 3D point cloud data, turnout, operating status, convolutional neural network, laser scanning

CLC Number:

WANG Dongyan, YU Cai, SHEN Kun, ZHANG Zhenjian, LI Yafeng. Application of point cloud data in geometric detection of critical nodes of turnouts[J]. Bulletin of Surveying and Mapping, 2024, 0(3): 107-112.

References

[1] YU Xueqiao,LANG Maoxiang,GAO Yang,et al. An empirical study on the design of China high-speed rail express train operation plan-from a sustainable transport perspective[J]. Sustainability,2018,10(7):2478.
[2] 王平,陈嵘,徐井芒,等. 高速铁路道岔系统理论与工程实践研究综述[J]. 西南交通大学学报,2016,51(2):357-372.
[3] 刘彦斌,彭强. 基于PC104的铁路道岔监测系统的设计与实现[J]. 铁路计算机应用,2007,16(8):20-22.
[4] 张宇宁,谢琦. 一种基于机器视觉的铁路道岔检测方法[J]. 计算机应用与软件,2015,32(1):225-228.
[5] MA Lingfei,LI Ying,LI J,et al. Mobile laser scanned point-clouds for road object detection and extraction:a review[J]. Remote Sensing,2018,10(10):1531.
[6] 何森,刘少丽,方玥,等. 铁路道岔场景识别与间距检测[J]. 计算机集成制造系统,2022,28(6):1823-1834.
[7] 张云鹏. 基于三维激光扫描技术的既有铁路道岔岔心自动提取方法[J]. 铁道勘察,2019,45(5):31-36.
[8] 李博闻,刘瑞,危凤海,等. 基于机载LiDAR的铁路工务设备及周边环境形变分析[J]. 铁道建筑,2023,63(1):143-147.
[9] 刘俊博,刘俊尧,孙淑杰,等. 基于激光点云的铁路边坡表面形变检测方法[J]. 铁道建筑,2021,61(11):82-85.
[10] 沈聪. 道岔综合监测技术的研究[D]. 北京:中国铁道科学研究院,2016:55-65.
[11] WANG Yue,SUN Yongbin,LIU Ziwei,et al. Dynamic graph CNN for learning on point clouds[J]. ACM Transactions on Graphics,38(5):146.
[12] GIRARDEAU-MONTAUT D. CloudCompare[D]. Paris:EDF R&D Telecom ParisTech,2016,11.
[13] CHARLES R Q,HAO Su,MO Kaichun,et al. PointNet:deep learning on point sets for 3D classification and segmentation[C]//Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Honolulu,HI:IEEE,2017:77-85.
[14] QI C R,YI Li,SU Hao,et al. PointNet++:deep hierarchical feature learning on point sets in a metric space[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems. Long Beach:ACM,2017:5105-5114.
[15] WIDYANINGRUM E,BAI Qian,FAJARI M K,et al. Airborne laser scanning point cloud classification using the DGCNN deep learning method[J]. Remote Sensing,2021,13(5):859.
[16] PIERDICCA R,PAOLANTI M,MATRONE F,et al. Point cloud semantic segmentation using a deep learning framework for cultural heritage[J]. Remote Sensing,2020,12(6):1005.