限界侵限检测方法研究与应用  

Research and Application of Detection Method of Boundary Intrusion

作　　者：赵鑫欣 孙淑杰 王凡 刘俊博 杜馨瑜 程雨 ZHAO Xinxin;SUN Shujie;WANG Fan;LIU Junbo;DU Xinyu;CHENG Yu(Infrastructure Inspection Research Institute,China Academy of Railway Sciences Corporation Limited,Beijing 100081,China)

机构地区：[1]中国铁道科学研究院集团有限公司基础设施检测研究所,北京100081

出　　处：《铁道建筑》2021年第3期155-158,共4页Railway Engineering

基　　金：中国铁道科学研究院集团有限公司基金(2020IMXM06);北京铁科英迈技术有限公司基金(2019IMAXM03)。

摘　　要：车载非接触式限界检测系统在获取线路及周边建筑物的空间位置信息时,安装在车头位置的高精度激光传感器容易受到异物和外部环境干扰。针对这一问题,本文提出了一种基于聚类分析和图像处理技术相结合的限界侵限检测方法。首先通过聚类分析预处理,将断面数据划分成不同的簇,再提取每个簇的轮廓特征,最后利用图像处理算法判定侵限状态。采用该方法对国内某线路进行了应用验证,结果表明,侵限检测结果平均准确率达到99%,比现有算法准确率提升了约15%,侵限分析处理时间小于4 ms,可以满足综合巡检车最高检测速度120 km/h的检测时效性和准确性。When the vehicle-mounted non-contact boundary detection system obtains the spatial position information of the railway line and surrounding buildings,the high-precision laser sensor installed in the front of the vehicle is vulnerable to the interference of foreign objects and external environment.In order to solve this problem,this paper proposed a detection method of boundary intrusion based on cluster analysis and image processing technology.Firstly,the cross-section data was divided into different clusters by cluster analysis preprocessing,and then the profile features of each cluster were extracted,finally,the image processing algorithm was used to determine the intrusion state.The method is applied to one domestic railway.The results show that the average accuracy of intrusion detection results is 99%,the accuracy is improved by about 15%by comparing with the existing algorithm.The time of intrusion analysis is less than 4 ms,which can meet the detection timeliness and accuracy of comprehensive inspection vehicle with the maximum detection speed of 120 km/h.

关 键 词：侵限检测 铁路建筑限界 试验研究 激光扫描 聚类分析 图像处理 算法 

分 类 号：U211.7[交通运输工程—道路与铁道工程] U216.3