Electrical Equipment Fault Diagnosis Based On Infrared Imaging Technology

With the constant expansion of China’s power grid as well as the rapiddevelopment of smart grid, It is very important for us to determine whether thefailure of electrical equipment. Infrared thermal imaging technology has anon-contact, no power outages, downtime advantages in electrical equipment faultdiagnosis, so it has been widely applied. However, the infrared images are subject tovarious noise (Electronic equipment itself, the surrounding environment) during thebuild process, so we need denoising. in the traditional manual inspection and imagemonitoring system, it require staff carefully checking and monitoring the occurrenceof overheating fault after the image is transferred to the monitor terminal, which willinevitably occur omissions or false, and waste a lot of manpower and resources, doesnot meet the modern smart grid construction requirements. So this paper presents theidea of electrical equipment fault automatic localization and diagnosis, by analyzingand studying infrared images or video, make sure the system automatically judgeelectrical equipment whether failure and automatically locate the fault location, thensend text messages and alarms.This paper briefly introduces the principle of overheating fault of electricalequipment, Since infrared images in general generate impulse noise and Gaussiannoise during the build process, this paper proposes the method which is pixel peergroups combined with the mean filter, the denoising significantly better in the case ofhigh-density impulse noise. Then the image’s edge detail is enhanced through NSCTtransform and achieve the infrared image preprocessing. Feature extraction is morecritical step in image processing. For infrared image uneven illumination, contrast isnot strong, the paper proposed the image segmentation method based on improvedPCNN, then achieve image recognition and classification after using fully affineinvariant ASIFT algorithm, ASIFT algorithm has a good affine results for image’svarious changes. Finally using the idea of topological matrix correction to automaticachieve fault-zone positioning, and then use the relative temperature differencemethod combined with temperature database to achieve electrical equipment faultdiagnosis.