基于卷积神经网络的输电杆塔缺陷智能识别系统研究与应用  

Research and Application of Transmission Tower Defect Intelligent Recognition System Based on Convolution Neural Network

作　　者：陈亮 原瀚杰 何勇 姚健安 董丽梦 CHEN Liang;YUAN Hanjie;HE Yong;YAO Jianan;DONG Limeng(Zhaoqing Power Supply Bureau,Guangdong Power Grid Co.,Ltd.,Zhaoqing Guangdong 526060,China)

机构地区：[1]广东电网有限责任公司肇庆供电局,广东肇庆526060

出　　处：《信息与电脑》2021年第1期142-144,共3页Information & Computer

摘　　要：针对当前电力系统输电线路的巡检现状,本文设计了一套贴合实际生产应用的输电缺陷智能识别系统,实现了数据快速分类、图像智能识别、识别结果人工审核、报告生成等功能,并且对两阶段的检测关键技术展开了研究。经过实践应用发现,该系统能够有效提高巡视数据分析效率,为输电线路实现智能化提供有效的支撑。Aiming at the current inspection status of power system transmission lines,this paper designs a set of transmission defect intelligent identification system suitable for actual production applications,which realizes the functions of rapid data classification,image intelligent identification,manual review of identification results,report generation,etc.The key technology of two-stage detection has been studied.Through practical application,it is found that the system can effectively improve the efficiency of patrol data analysis and provide effective support for the realization of intelligent transmission lines.

关 键 词：卷积神经网络 杆塔缺陷 智能识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TM75[自动化与计算机技术—计算机科学与技术]