高速铁路接触网悬挂装置开口销不良状态检测方法研究  

Defective Condition Detection of Split Pins in Catenary Suspension Device of High-speed Railway

作　　者：钟俊平[1] 刘志刚[1] 陈隽文 韩志伟[1] ZHONG Junping;LIU Zhigang;CHEN Junwen;HAN Zhiwei(School of Electrical Engineering,Southwest Jiaotong University,Chengdu G10031,China)

机构地区：[1]西南交通大学电气工程学院,四川成都610031

出　　处：《铁道学报》2018年第6期51-59,共9页Journal of the China Railway Society

基　　金：国家自然科学基金(U1434203;51377136;51407147);四川省青年科技创新团队项目(2016TD0012)

摘　　要：针对高速铁路接触网支撑装置中开口销缺失、松脱和张角不足问题,提出两种基于灰度分布特征的检测判据。在定位阶段,依据斜腕臂及其连接处的线特征,对斜腕臂进行Hough直线检测并提取其两侧边缘的斜率特征,实现斜腕处开口销的初定位,利用PBoW模型对初定位图片进行分类;采用SIFT算法准确定位非斜腕处开口销。在检测阶段,使用Hough累加矩阵提取连接处螺钉的圆形特征,对销钉所处的圆环区域实现精确提取;针对圆环内无其他零部件遮挡和干扰情况,根据圆环内非连通区域分布规律给出表面开口销缺失和张角不足检测判据;根据销钉受力部分和两端非受力部分的灰度分布,给出双耳下方开口销不良状态检测判据。实验表明,本文方法能够较准确地实现多位置开口销不良状态的同时检测。Targeting the missing,loosening and opening angle insufficiency of split pins in the catenary support devices of high-speed railway,two detective criteria based on grayscale distribution characteristics were put forward in this paper.In the positioning stage,Hough transform was adopted to process the image on the basis of line features on the oblique bracket and junction,whose upper and lower edge slope features were extracted respectively,to realize the preliminary location of split pins along the oblique bracket.Besides,the Pyramid Bag of Word model was adopted to classify the preliminary location images.And then the SIFT algorithm was used to locate the pins that were not along the oblique bracket.In the period of detection,the Hough accumulative matrix was used to extract circular features on the joint screw and to achieve the accurate extraction of the ring region where the split pin lies.With no occlusion or interference of other parts,the detective criterion for the insufficient opening angle and the lack of the pin on the surface of the part was given according to the distribution characteristics of the disconnected regions in circular ring.The detective criterion for the pins under the clevis was given according to the gray-scale distribution of the force part and non-force parts on both sides of the pin.The experiments show that the method can realize defective condition detection of split pins on multiple locations accurately and simultaneously.

关 键 词：开口销不良状态 HOUGH变换 PBoW模型分类 SIFT算法 灰度分布特征 

分 类 号：U225[交通运输工程—道路与铁道工程]