基于改进SVM的变电站设备红外图像故障诊断方法研究  

Research on infrared image fault diagnosis method for substation equipment based on improved SVM

作　　者：郑祥 张鸿鹄 魏岩岩 ZHENG Xiang;ZHANG Honghu;WEI Yanyan(Ultra High Voltage Branch,State Grid Anhui Electric Power Co.,Ltd.,Hefei 230000,China)

机构地区：[1]国网安徽省电力有限公司超高压分公司,合肥230000

出　　处：《电测与仪表》2024年第12期49-55,共7页Electrical Measurement & Instrumentation

基　　金：国家电网有限公司科技项目(52120320006T)。

摘　　要：针对现有智能变电站电压致热型设备红外图像故障诊断方法存在的诊断准确率差和效率低等问题,提出一种多特征融合方法和改进支持向量机相结合的智能变电站电压致热型设备红外图像故障诊断方法。通过多特征融合方法将颜色特征、边缘特征、纹理特征进行融合,通过改进的帝国竞争算法对支持向量机参数(惩罚因子和核参数)进行优化,提高故障诊断性能。通过算例进行比较分析,验证所提方法的可行性。结果表明,所提方法在多个电压致热型设备故障诊断中具有较高的故障诊断性能,优于单一特征故障诊断方法,故障诊断准确率为94.83%,平均诊断时间为0.62 s,为无人变电站的发展奠定了基础。A method for infrared image fault diagnosis of voltage heating equipment in intelligent substations is proposed,which combines multiple feature fusion methods with improved support vector machine(SVM),to address the issues of poor diagnostic accuracy and low efficiency in existing infrared image fault diagnosis methods for voltage heating equipment in intelligent substations.By using multiple feature fusion methods to fuse color features,edge features,and texture features,and optimizing SVM parameters(penalty factors and kernel parameters)through an improved imperial competition algorithm,and the fault diagnosis performance is improved.The feasibility of the proposed method is verified by comparing and analyzing numerical examples.The results show that the proposed method has high fault diagnosis performance in multiple voltage induced thermal equipment fault diagnosis,superior to the single feature fault diagnosis method,with a fault diagnosis accuracy rate of 94.83%and an average diagnosis time of 0.62 seconds.This has laid the foundation for the development of unmanned substations.

关 键 词：智能变电站 红外图像 故障诊断 电压致热型设备 支持向量机 

分 类 号：TM732[电气工程—电力系统及自动化]