基于改进深度学习的电表终端故障图像识别方法  

Fault Imager Recognition Method of Meter Terminal Based on Improved Deep Learning

作　　者：丁超 张秋雁 胡厚鹏 张俊玮 欧家祥 DING Chao;ZHANG Qiuyan;HU Houpeng;ZHANG Junwei;OU Jiaxiang(Electric Power Research Institute of Guizhou Power Grid Co.,Ltd.,Guiyang 562400,China)

机构地区：[1]贵州电网有限责任公司电力科学研究院,贵州贵阳562400

出　　处：《电工技术》2020年第10期36-39,共4页Electric Engineering

摘　　要：针对常规运维模式越来越难以满足电表终端高效可靠运营需求的问题,文章提出了一种基于改进深度学习的电表终端故障图像识别方法。在分析现有基于深度学习电表故障识别方法的不足的基础上,介绍了改进深度学习识别方法的思路。进一步设计了改进深度学习识别方法,具体阐释了深度学习分类网络、电表终端检测网络、组态匹配三个主要技术环节。最后以一个算例,介绍了改进深度学习的电表终端故障图像识别方法的应用情况,证明了该方法的有效性。In view of the problem that the conventional operation and maintenance mode are more and more difficult to meet the demand of efficient and reliable operation of the meter terminal,a fault image recognition method of the meter terminal based on improved deep learning was proposed.Based on the analysis of the shortcomings of the existing methods of fault identification of meters based on deep learning,the ideas of improving the methods of fault identification of meters based on deep learning was introduced.Furthermore,the improved deep learning recognition method was designed.The three main technical links of deep learning classification network,meter terminal detection network and configuration matching were explained.Finally,an example was given to show the application of the improved in-depth learning method for fault image recognition of ammeter terminals,which proved the effectiveness of the method.

关 键 词：电表终端 深度学习 人工智能 图像识别 故障识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]