基于自适应支持向量机的变压器温度异常检测方法  

Transformer Temperature Anomaly Detection Method Based on Adaptive Support Vector Machine

作　　者：郭成英 韩玉 陶梦江 郭大勇 樊磊 GUO Cheng-ying;HAN Yu;TAO Meng-jiang;GUO Da-yong;FAN Lei(State Grid Anhui Electric Power Co.,Ltd.,Hefei 230000;State Grid Anhui Ultrahigh Voltage Company,Hefei 230000)

机构地区：[1]国网安徽省电力有限公司,合肥230000 [2]国网安徽电力有限公司超高压分公司,合肥230000

出　　处：《环境技术》2024年第4期77-83,共7页Environmental Technology

摘　　要：由于忽略了环境温差的影响,电力变压器表面温度检测结果的残差值较大。因此,基于自适应支持向量机的变压器温度异常检测方法。采集并增强、预处理电力变压器表面温度红外图像,采用添加了高斯核的自适应支持向量机,初步分类红外图像,构建高斯核特征空间,分类线性特征与非线性特征;考虑环境温度的影响,分析电力变压器的常规表面温度,经过差值运算后,结合自适应运算过程,得到温度异常检测结果。实验结果表明:所提方法应用后,电力变压器表面温度检测结果的温度残差值较小,检测准确度较高,满足了电力变压器的安全运行需求。Due to the neglect of the influence of environmental temperature differences,the residual value of the surface temperature detection results of power transformers is relatively large.Therefore,a transformer temperature anomaly detection method based on adaptive support vector machine.Collect,enhance,and preprocess infrared images of the surface temperature of power transformers.Using an adaptive support vector machine with Gaussian kernel added,preliminarily classify the infrared images,construct a Gaussian kernel feature space,and classify linear and nonlinear features;Considering the influence of environmental temperature,the conventional surface temperature of power transformers is analyzed.After difference calculation,combined with adaptive calculation process,temperature anomaly detection results are obtained.The experimental results show that after the application of the proposed method,the temperature residual value of the surface temperature detection results of power transformers is small,the detection accuracy is high,and it meets the safe operation requirements of power transformers.

关 键 词：电力变压器:变压器异常 自适应支持向量机 红外图像 表面温度 异常检测 

分 类 号：TM407[电气工程—电器]