基于二代曲波系数形态学条带能量法的绝缘子故障检测  

Insulator Fault Detection Based on Curvelet Coefficients Morphology and Zonal Energy Method

作　　者：韩志伟[1] 刘志刚[1] 杨红梅[1] 韩烨[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》2013年第3期36-40,共5页Journal of the China Railway Society

基　　金：国家自然科学基金(U1134205;51007074);教育部新世纪优秀人才支持计划(NCET-08-0825);铁道部科技研究开发计划(2011J016-B);中央高校基本科研业务费专项资金资助项目(SWJTU11CX141)

摘　　要：针对电气化铁路接触网绝缘子瓷瓶破损故障检测问题,提出一种基于智能图像处理技术的故障识别算法。利用二代曲波变换对综合检测列车现场拍摄图像进行各向异性方向滤波获取绝缘子特征量;为解决绝缘子不同瓷瓶特征系数分布不规则问题,利用方向性形态学闭运算对曲波系数均衡化;利用能量条带法对曲波系数矩阵进行降维,最终获得绝缘子位置信息,判断绝缘子是否存在破损故障。实验表明:该方法可直接对现场拍摄的图像进行全局分析,对于图像中绝缘子定位以及绝缘子故障信息的判断都达到较高准确性,为电气化铁路绝缘可靠性检测提供一种思路。The fault recognition algorithm based on the intelligent image processing theory was proposed to de tect breakages of porcelain insulators of catenaries of electrified railways. Characteristics of insulators were filtered out by Secondary Generation Curvelet Transform in the anisotropic direction from the images shot on site by use of train inspection car. A new curvelet coefficients adjustment algorithm was used to balance the distribution of different ceramic chips by using the closing operation of the directional mathematical morphology. The zonal energy method was proposed to reduce the dimension of the curvelet matrix and simplify the analysis. The positions of insulators were identified and breakage faults were recognized finally. The experiments show that the proposed method facilitates direct and overall analysis on site images and realizes highly accurate judgment of locations and breakages of insulators, thus providing a new way of thinking to detection of insulation reliability of electrified railways.

关 键 词：图像处理 绝缘子 曲波 形态学 故障检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]