基于图像识别的输电线路设备缺陷识别应用系统设计  

Design of transmission line equipment defect recognition application system based on image recognition

作　　者：翟瑞聪 林俊省 郑桦 ZHAI Ruicong;LIN Junsheng;ZHENG Hua(Machine Patrol Management Center of Guangdong Power Grid Co.,Ltd.,Guangzhou 510000,China;Digital Grid Research Institute Co.,Ltd.,China Southern Power Grid,Guangzhou 510000,China)

机构地区：[1]广东电网有限责任公司机巡管理中心,广东广州510000 [2]南方电网数字电网研究院有限公司,广东广州510000

出　　处：《电子设计工程》2022年第6期161-164,169,共5页Electronic Design Engineering

摘　　要：为更好地保障输电线路设备的电子汇集能力,并实现对高压电网络结构的合理性规划,基于图像识别技术设计了一种输电线路设备缺陷识别应用系统。联合输电循环回路,妥善连接任务管理模块与电量基站控制设备,完成对系统硬件执行环境的搭建。在此基础上,设计卷积神经网络,借助栈式自动编码器结构体,对RBM识别节点进行训练处理,完成系统软件执行环境的搭建。将软硬件结构结合,实现基于图像识别输电线路设备缺陷识别系统的顺利应用。对比实验结果表明,与传统的深度学习型缺陷识别系统相比,该文系统可同时检测的传输电子量更多,且既定时间节点处的识别精准度也更高,能够较好地实现对输电线路设备电子汇集能力的保护。In order to better guarantee the electronic gathering ability of transmission line equipment and realize the rational planning of the high-voltage network structure,an application system of transmission line equipment defect recognition based on image recognition technology is designed.The combined transmission cycle loop can properly connect the task management module and the power base station control equipment to complete the construction of the system hardware execution environment.On this basis,a convolutional neural network is designed.With the help of the stack automatic encoder structure,the RBM recognition nodes are trained and processed to complete the construction of the system software execution environment.The software and hardware structures are combined to realize the smooth application of the transmission line equipment defect recognition system based on image recognition. The experimental results show that compared with the traditional deep learning defect recognition system,the system in this paper can simultaneously detect more transmitted electrons,and the recognition accuracy at the fixed time node is also higher,which can better realize the protection of the electronic collection ability of transmission line equipment.

关 键 词：图像识别 缺陷识别 循环回路 卷积神经网络 自动编码器 RBM节点 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]