高速铁路接触网承力索座识别与辅助承力索缺失故障检测方法研究  

Loss Fault Detection for Auxiliary Catenary Wire of High-speed Railway Catenary Wire Holder

作　　者：张桂南[1] 刘志刚[1] 韩烨[1] 韩志伟[1] 

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》2017年第5期40-46,共7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承力索座辅助承力索缺失的问题,基于DHOG特征及离散余弦变换特征增强提出辅助承力索缺失故障检测方法。该方法首先需对样本库图像提取DHOG特征,在此基础上训练AdaBoost分类器并给出承力索座的精确定位;其次使用离散余弦变换滤除图像背景信息,实现辅助承力索目标特征信息增强;随后通过可接受圆弧检测、圆弧段聚类拟合实现图像内的圆形检测和统计,并根据统计数目,给出故障判据,最终实现高铁接触网辅助承力索缺失的故障判断。实验结果表明本文方法在目标定位上具有较好的尺度、旋转不变性,在故障检测上具有较高的检测精度。To deal with the fault of the auxiliary catenary wire loss of high-speed railway catenary wire holder, a method to detect the loss fault of the auxiliary catenary wire based on DHOG features and discrete cosine transform feature enhancement was proposed. Firstly, with training AdaBoost classifier, precise positioning of the catenary wire holder was given based on the DHOG feature extraction of the image sample library. Secondly, the discrete cosine transform was used to filter out the background information of the image to realize the enhancement of the target feature information of the auxiliary catenary wire. Thirdly, with the acceptance of circular arc detection and the fitting of arc segment clustering, the circular detection and statistics within the image can be realized. Based on the statistics, the failure criterion was given. Finally, the fault judgment of the auxiliary catenary wire loss was realized. The experimental results showed that the proposed method in this paper has good scale and rotation invariance in terms of target location, with high detection accuracy for the auxiliary catenary.

关 键 词：辅助承力索 DHOG特征 ADABOOST分类器 离散余弦变换 圆弧分段拟合 

分 类 号：U226.81[交通运输工程—道路与铁道工程]