基于智能感知与深度学习的智能变电站设备状态检测方法  

State Detection Method of Smart Substation Equipment Based on Intelligent Perception and Deep Learning

作　　者：李远松 丁津津 徐晨 高博 汤汉松 单荣荣[4] LI Yuansong;DING Jinjin;XU Chen;GAO Bo;TANG Hansong;SHAN Rongrong(State Grid Anhui Electric Power Research Institute,Hefei 230601;Anhui Xinli Electric Technology Consulting Co.,Ltd.,Hefei 230022;Jiangsu LingChuang Electric Automation Co.,Ltd.,Zhenjiang 212009;NARI Technology Development Co.,Ltd.,Nanjing 211106)

机构地区：[1]国网安徽省电力有限公司电力科学研究院,合肥230601 [2]安徽新力电业科技咨询有限责任公司,合肥230022 [3]江苏凌创电气自动化股份有限公司,镇江212009 [4]国电南瑞科技股份有限公司,南京211106

出　　处：《电气工程学报》2022年第2期208-214,共7页Journal of Electrical Engineering

基　　金：国家重点研发计划资助项目(2018YFB0905000)。

摘　　要：针对现有变电站设备状态检测方式单一、检测效果欠佳的问题,提出一种基于智能感知与深度学习的智能变电站设备状态检测方法。首先,在变电站四个角落安装低功率的热像仪,以实时监测设备状态。然后,应用中值滤波和侵蚀技术处理设备热图像,获得灰度图像后,利用加速鲁棒特征法提取图像特征,初步监测设备状态。最后,基于深度学习模型对图像特征作训练分类,以检测存在故障的设备。基于Tensorflow平台对其性能进行试验论证,结果表明,相比于其他方法,所提方法的检测准确率和召回率更高且检测速率更快,能够直观准确地掌握变电站的设备状态。Aiming at the problems of single state detection mode and poor detection effect of existing substation equipment,a state detection method of Smart substation equipment based on intelligent perception and deep learning is proposed.Firstly,low-power thermal imagers are installed in four corners of the substation to monitor the status of the equipment in real time.Then,the thermal image of the equipment is processed by median filter and erosion technology.After getting the gray image,the image features are extracted by accelerated robust feature method,and the status of the equipment is preliminarily monitored.Finally,the image features are further analyzed based on the deep learning model to detect the faulty equipment.The proposed method is based on Tensorflow platform to demonstrate its performance.The results show that compared with other methods,the proposed method has higher detection accuracy and recall rate,and faster detection rate,which can intuitively and accurately grasp the equipment status of the substation.

关 键 词：智能变电站 深度学习 智能感知 中值滤波 加速鲁棒特征法 

分 类 号：TM73[电气工程—电力系统及自动化]