摘要
为对变电站设备进行实时监测,设计一种基于热像仪的变电站设备状态监测系统,并提出一种利用热像仪获取热像图的变电站设备状态监测算法。首先,热像仪采集设备的热像图和正常图像,并应用光学字符识别方法找出图像中的最高温度和最低温度;其次,应用中值滤波和腐蚀技术得到矩形框为界的可能断层区域;再根据矩形框裁剪图像提取裁剪图像的加速鲁棒特征;最后,通过训练好的自适应神经模糊推理系统(adaptive neuro fuzzy inference system,ANFIS)或支持向量机(support vector machine,SVM)分类器对提取的加速鲁棒特征进行高级阈值处理,从而识别变电站设备状态。实验结果表明,所提算法采用SVM能够正确识别变电站设备临界故障状态,而算法采用ANFIS识别变电站设备临界故障状态还存在一定错误。
For the real-time monitoring of substation equipment,a condition monitoring system of substation equipment based on thermal cameras is designed,and a novel condition monitoring algorithm of substation equipment is proposed by using thermal images obtained from thermal cameras.Firstly,the thermal image and normal image of the equipment are collected by thermal cameras,and the maximum and minimum temperatures in the image are found by using the optical character recognition method.Secondly,the possible fault area bounded by rectangle frame is obtained by median filter and corrosion technology.Thirdly,according to the rectangular frame clipping image,the speeded-up robust feature of the clipping image is extracted.Finally,the state of substation equipment can be identified by using the trained ANFIS or SVM classifier to process the speeded-up robust features with advanced threshold value.Experimental results show that the proposed algorithm can correctly identify the critical fault conditions of substation equipment by using SVM,while there are some errors in the algorithm using ANFIS to identify the critical fault conditions of substation equipment.
作者
李文震
郭海波
朱吕甫
LI Wenzhen;GUO Haibo;ZHU Lvfu(Equipment Management Department,State Grid Inner Mongolia East Power Co.,Ltd.,Hohhot 010000,China;Tongliao Power Supply Company,State Grid Inner Mongolia Eastern Electric Power Co.,Ltd.,Tongliao 028000,China;Anhui Jushi Technology Co.,Ltd.,Hefei 230000,China)
出处
《电力信息与通信技术》
2022年第3期48-57,共10页
Electric Power Information and Communication Technology
关键词
支持向量机
自适应神经模糊推理系统
变电站
热像仪
电网
support vector machine
adaptive neuro fuzzy inference system
substation
thermal camera
power grid
