继保压板状态图像识别与核对系统设计和实现  

Design and Implementation of State Image Recognition and Verification System for Relay Protection Plate

作　　者：侯伟 李新海 范德和 肖星 梁景明 林悦德 HOU Wei;LI Xinhai;FAN Dehe;XIAO Xing;LIANG Jingming;LIN Yuede(Zhongshan Power Supply Bureau of Guangdong Power Grid Co.,Ltd.,Zhongshan 528400,China)

机构地区：[1]广东电网有限责任公司中山供电局,广东中山528400

出　　处：《电工技术》2021年第24期63-66,共4页Electric Engineering

基　　金：广东电网有限责任公司科技项目(编号GDKJXM20184171(032000KK52180060))。

摘　　要：变电站人工核对压板工作量大、效率低,造成不能及时发现压板误投退、漏投退隐患,因此提出了一种以机器人和手持终端为底层感知设备的继保压板状态图像识别与核对系统。该系统采用多层设计架构方式,实现了设备的实时控制和数据的高效传输与处理。系统感知层采用轮式机器人和手持终端相结合的方式对压板屏柜进行巡检,服务器端采用YOLO目标识别算法实现压板投退状态的识别,使用SSM框架快速搭建稳定性良好的应用系统。该系统在某220 kV变电站进行了试点应用,对该站主控室45面压板保护屏柜(共计1107个压板)进行每月1次定时巡检,并自动形成压板核对报告,压板投退状态识别正确率达100%,完全满足压板日常巡检要求,运行效果良好。Manual checking of pressure plates in substations is a big workload and low efficiency,resulting in the failure to find the pressure plates in time and the hidden danger of missing and omitting to throw them back.Therefore,a protection plate status image recognition and verification system based on robot man and handheld terminal as the underlying sensing device is designed.The system design adopts a multi-layer design architecture approach to achieve real-time control of equipment and efficient transmission and processing of data.The system perception layer uses a combination of wheeled robots and handheld terminals to inspect the pressure plate screen cabinets.The server side uses YOLO target recognition algorithm to realize the recognition of pressure plate throwback status,and uses the SSM framework to quickly build a well-stabilized application system.The system is piloted in a 220 kV substation.45 plate protection panel cabinets(a total of 1107 plates)in the main control room of the station are inspected regularly once a month,and the plate check report is formed automatically.The correct identification rate of the plate input and output status reached 100%,which fully met the requirements of the plate daily inspection and had good operation effect.

关 键 词：继电保护压板 图像识别 自动核对系统 YOLO SPRING框架 VueJS 

分 类 号：TM58[电气工程—电器]