Research On Infrared Image Analysis Method Of Electrical Equipment Based On Deep Learning

In order to improve the intelligent level of electrical equipment condition monitoring,many monitoring technologies have been actively promoted and applied.Among many monitoring technologies,infrared thermal imaging technology is favored for its advantages of no power failure,no sampling and no disintegration.With the increasing use of infrared diagnostic technology in power systems,the number of infrared imagers and infrared images taken during inspections have also increased year by year.At present,the analysis and diagnosis of infrared images are mostly dependent on labor.However,the current infrared monitors are less able to satisfy the analysis of huge amounts of infrared images,and the accumulation of professional knowledge and monitoring experience of some monitors is insufficient.The ability to analyze and judge is poor,which greatly restricts the improvement of the intelligent level of condition monitoring.With the development of deep learning in the field of computer vision,this paper introduces deep learning into the analysis and processing of infrared images of electrical equipment.Firstly,this paper introduces the traditional infrared image processing technology and proposes its shortcomings.Then it proposes more intelligent deep learning.It introduces the theoretical basis of deep learning,its application in image processing and its theory of migration learning.Then,the deeper learning image segmentation application is studied in more depth.From AlexNet,VGG,ResNet to FCN and Mask R-CNN,the theoretical basis and application effects of deep learning in image segmentation are analyzed.Finally,using Mask R-CNN and migration learning to detect and segment the infrared image of electrical equipment,the experimental environment of training,the structure of training network,training parameters are introduced in detail,and the network is trained in the infrared image dataset of electrical equipment.The effect of different iterations on the model results shows that the proposed method can achieve the target detection and segmentation of electrical equipment in infrared images.