基于机器学习的电力设备故障红外智能诊断方法  

Infrared intelligent diagnosis method of power equipment fault based on machine learning

作　　者：托娅 王伟 毛华敏 程宏波 王林 TUO Ya;WANG Wei;MAO Huamin;CHENG Hongbo;WANG Lin(Inner Mongolia Electric Power Group Co.,Ltd.,Hohhot 010000,Inner Mongolia,China;School of Electrical and Automation Engineering,East China Jiaotong University,Nanchang 330013,Jiangxi,China)

机构地区：[1]内蒙古电力(集团)有限责任公司,内蒙古呼和浩特010000 [2]华东交通大学电气与自动化工程学院,江西南昌330013

出　　处：《河南理工大学学报（自然科学版）》2022年第5期121-126,共6页Journal of Henan Polytechnic University（Natural Science）

基　　金：国家自然科学基金资助项目(51967007);江西省重点研发计划项目(20202BBEL53008);江西省杰出青年人才培养项目(20162BCB23046)。

摘　　要：针对电力设备红外检测诊断方法落后、效率低等问题,采用双层网络进行设备类型识别和结构区域划分,从而实现快速有效的诊断。首先利用R-FCN建立电力设备识别模型,利用Mask RCNN实现电力设备区域结构的识别结果自动分割,根据划分结构自动提取不同区域的最高温度,依据识别设备类型调用不同判据自动诊断设备状态。搭建红外智能诊断平台进行实验,结果表明:该方法识别准确率高、状态判断结果可靠,无需大量的故障样本,可为电力设备故障的红外智能诊断提供一种快速有效的处理方法。Aiming at the problems of backward method and low efficiency of the power equipment infrared detection and diagnosis,the double-layer network was used to identify the equipment type and to divide the equipment structure,so as to realize the fast and effect diagnosis.Firstly,the R-FCN was used to establish the identification model of electrical equipment,and the Mask RCNN was used to segment the regional structure of power equipment.The maximum temperature of different regions was extracted automatically according to the divided structure,then the equipment status could be diagnosed by criteria according to the identified equipment type.The infrared intelligent diagnosis platform was built,and the experimental results showed that the method had high recognition accuracy,reliable state judgment results,moreover it did not need a large number of fault samples,which provided a fast and effective method for infrared diagnosis of electrical equipment.

关 键 词：电力设备 红外检测 深度学习 智能诊断 

分 类 号：TP206.3[自动化与计算机技术—检测技术与自动化装置]