| With the rapid development of national science and the rapid improvement of the economic level,the scale of the power grid is expanding,and the load is increasing accordingly.In order to cope with the huge load,the number of substations is also increasing in recent years,and the tasks undertaken by the limited power grid operators are also increasing.In order to reduce the pressure of operation and maintenance and eliminate human omissions,the intelligence and intelligence of the power grid have become the development trend in recent years.However,most of the intelligent tasks are focused on primary equipment,and less attention is paid to other parts.In the actual daily operation and maintenance,the inspection work in the substation control room,communication room,main control room and other computer rooms is also part of the inspection.In order to improve the efficiency of substation inspection tasks,this paper focuses on artificial intelligence technology in smart substations Research on the application of computer room monitoring system,The main contents include:(1)The preprocessing of the pictures collected by the lens is studied,including image distortion and image matching and stitching.In the preprocessing of image distortion,use the inspection camera to take a picture of a standard calibration checkerboard,extract the corner points,and then based on the corner points,solve the internal parameter matrix and external parameter matrix of the lens and optimize the estimation,complete the camera calibration,remove the effect of image distortion.In the preprocessing of image matching and stitching,the SURF operator is used to extract the feature points in the picture,and then the RANSAC algorithm is used to match and stitch multiple pictures.(2)The identification of indicator lights was studied.First,filter the collected pictures,and then convert them from RGB color space to HSV color space.By adjusting the threshold,divide the lights of different colors in the H channel,and divide the brightness and darkness of the lights in the V channel.Finally,use the Hough circle detection method locates the position of the indicator light.(3)The recognition of screen text is studied.Comparing the traditional text recognition method and several methods based on deep learning,finally selected the Convolutional Recurrent Neural Network algorithm,collected a large number of pictures as a training set,trained it and got good results.(4)Design the intelligent monitoring system,including the hardware system,software function and operation process,obtain the image data through the camera equipped on the robot,and then provide intelligent analysis of the intelligent monitoring system during the operation and maintenance process of the computer room through the artificial intelligence platform Serve. |
