Research On Equipment Condition Monitoring Method Based On Infrared Image Of Substation Scene

The substation is an important part of the power grid transmission and distribution system,which directly affects the safe and stable operation of the power grid.Therefore,building a real-time and accurate substation equipment condition monitoring system is an important part of power grid management.Infrared detection technology can detect the operating status of electrical equipment non-destructively by detecting the surface temperature of the equipment,and different equipment corresponds to different normal temperatures,so the key to equipment status detection is to be able to accurately identify the type of equipment and find out the infrared image of each equipment Temperature measurement area.In this paper,the infrared image of the equipment taken by a substation in Yunnan Power Grid is taken as the research object,and the equipment state monitoring method of the substation scene is studied.The main contents include:Based on image quality evaluation indicators such as image entropy,average gradient and number of feature points,the characteristics of infrared images in substation scenes are analyzed.In view of the shortcomings of low spatial resolution of infrared images and the small number of feature points,combined with two kinds of infrared image enhancement algorithms,such as gamma correction with improved nonlinear mapping relationship and subblock overlapping histogram equalization,to achieve the premise of maintaining the original image structure,Make full use of local details to enhance image quality.Aiming at the problem of accuracy and lack of real-time recognition of traditional methods for identifying electrical equipment types,this paper proposes a target detection-based substation equipment status detection model.By adjusting the loss function and network structure,the model detection accuracy and training speed are improved.The model is used to obtain the device type information and the pixel distribution of the corresponding image.The infrared temperature of the area is analyzed to obtain the device temperature and compared with the normal temperature of the device to obtain the device operating state.The verification results show that the model truly reflects the operating state of the equipment within the allowable error range.In order to facilitate the observation of the overall status distribution information of the substation,this article improves the three parts of feature extraction,image matching and image fusion based on the traditional image stitching method.Through example verification,it can be seen that the feature matching time is reduced after the improvement compared to the algorithm before the improvement Small,better stitching effect.Extension to the stitching of multiple small parallax scene images can obtain good visual effects and achieve accurate monitoring of the status of substation equipment.