基于图像识别的10 kV电缆接头施工缺陷算法应用研究  

Application Research of 10 kV Cable Joint Construction Defect Algorithm Based on Image Recognition

作　　者：薛艺为 党卫军 孙奇珍 霍建彬 方春华[2] XUE Yiwei;DANG Weijun;SUN Qizhen;HUO Jianbin;FANG Chunhua(Huangpu Bureau of Guangzhou Electric Power Company Huangpu District,Guangzhou 510700,China;China Three Gorges University,Hubei Yichang 443000,China)

机构地区：[1]广州电力公司黄埔局,广州510700 [2]三峡大学,湖北宜昌443000

出　　处：《电工材料》2021年第6期20-24,共5页Electrical Engineering Materials

摘　　要：在研究10 kV电力电缆中间接头的典型缺陷类型时,通过分析施工过程,得出中间接头典型施工缺陷有主绝缘表面污渍、主绝缘划伤、外半导电层剥离不齐3类。根据污渍与主绝缘色差大,提出采用灰度色差作为判定缺陷的有效特征量;针对划伤缺陷区域与主绝缘灰度色差不明显的问题,通过提取纹理特征,达到有效区分主绝缘污渍与划伤缺陷的目的。对于主绝缘缺陷,采用基于神经网络的缺陷识别方法,能有效识别出主绝缘上的污渍与划伤两类缺陷。应用以上技术手段,实现了对整个电缆中间接头施工缺陷的判定,整体检测的有效性有较大的提高,证实了针对性算法的有效性。In the study of the typical defect types of 10 kV power cable intermediate joints,the construction process was analyzed,and the typical construction defects of the intermediate joints were found to be main insulation surface stains,main insulation scratches,and outer semi-conductive layer peeling.According to the large difference between the stain and the main insulation,the gray color difference is used as the effective feature quantity to determine the defect.For the problem that the scratch defect area and the main insulation gray level difference are not obvious,the texture feature is extracted to achieve the effective distinction between the main insulation stain and the stroke.The purpose of the injury.For the main insulation defect,the neural network-based defect recognition method is used to effectively identify the two types of defects such as stain and scratch on the main insulation.The determination of the construction defects of the entire cable intermediate joint is realized,and the overall detection is greatly improved,thereby confirming the effectiveness of the targeted algorithm.

关 键 词：10 kV电缆 中间接头 主绝缘 半导电层 典型缺陷 边缘检测 

分 类 号：TM835[电气工程—高电压与绝缘技术]