摘要
传统红外人工诊断方法难以应对变电站机器人、无人机自主巡检产生的海量红外图片,目前针对电流致热型缺陷较易识别,但缺少危害严重的电压致热型缺陷智能诊断方法研究,提出了一种基于旋转目标检测的变电设备电压致热型缺陷智能诊断方法。基于改进R^(3)Det模型对瓷套进行旋转目标检测,基于Faster RCNN模型对红外图像中三相区域、套管、电流互感器等变电设备区域进行识别、定位;通过自动关联包含在三相区域中的同类设备,计算同类设备温差;基于温差阈值法进行电压致热型缺陷诊断。使用现场采集红外图像进行训练和测试,结果表明:目标检测平均精度均值为90.65%,电压过热型缺陷识别准确率达到81.39%,误报率为9.62%,实验结果证明所提方法能够有效地从红外图像中自动识别电压致热型缺陷,可为实现机器巡检作业红外诊断智能化奠定基础。
It is difficult to analyze and process massive infrared images from autonomous inspection of substation robots and unmanned aerial ve-hicles(UAVs) with traditional manual diagnosis method. At present, it is easy to identify current induced defects, whereas, the researches on intelligent diagnosis methods for voltage induced defects are rarely available,therefore, an intelligent diagnosis method for voltage induced thermal defects of substation equipment based on rotating target detection is proposed in this paper. The rotating target of porcelain bushing is detected based on the improved R^(3)Det model. The three-phase area, bushing, current transformer, and other substation equipment areas in infrared image are identified and located based on the Fast RCNN model;the temperature difference of the same kind of equipment is calculated by automatically associating the same equipment contained in the three-phase region;based on the temperature difference threshold method, the voltage induced defects were diagnosed. Infrared images on the spot were collected for training and testing. The results show that the average accuracy of target detection is 90.65%, the recognition accuracy of voltage type thermal defects is 81.39%, and the false alarm rate is 9.62%. The experimental results show that the proposed method can be adopted to automatically diagnose voltage induced thermal defects from infrared images, which lays a foundation for intelligent infrared diagnosis of machine inspection.
作者
李文璞
毛颖科
廖逍
谢可
刘迪
张晓航
LI Wenpu;MAO Yingke;LIAO Xiao;XIE Ke;LIU Di;ZHANG Xiaohang(State Grid Information&Telecommunication Group Co.,Ltd.,Beijing 102211,China;Maintenance Branch,State Grid Shangai Electric Power Company,Shanghai 200063,China)
出处
《高电压技术》
EI
CAS
CSCD
北大核心
2021年第9期3246-3253,共8页
High Voltage Engineering
基金
国家电网公司总部科技项目(5500-202017083A-0-0-00)。
关键词
红外图像
缺陷识别
变电设备
旋转目标检测
R^(3)Det
智能诊断
infrared image
defect recognition
substation equipment
rotating target detection
R^(3)Det
intelligent diagnosis
