Application Of Image Processing In Substation Pointer Instrument Identification

With the advancement of science and the investment in smart grids,inspection robots have gradually been applied to the detection and identification of pointer meters in substations.However,at present,the automatic inspection of the instrument still needs human participation to read instrument images.Therefore,the goal of the intelligent substation cannot be truly realized.In addition,the reading results of the instrument will also be greatly affected by the shooting quality of the inspection robot.In this regard,this paper proposes a substation instrument recognition system based on image processing.The instrument detection,instrument image pre-processing and instrument automatic reading algorithm were systematically studied,and corresponding improvements were made to realize highly applicable substation instrument identification.The main work of this paper is as follows:1.Firstly,in order to solve the problems such as randomness of instrument position,environmental complexity and multi-type instruments,this paper proposes the system design framework and technical solutions based on instrument detection,instrument pre-processing and instrument identification,which reduces the difficulty to recognize.2.Secondly,this paper proposes an improved instrument detection algorithm based on Faster R-CNN.On the basis of the original Faster R-CNN algorithm,we use multi-scale training,increased anchor number and online hard example mining strategy to enhance the performance of the model.The algorithm can accurately detect the target instrument from the complex background.At the same time,the center position magnified high quality instrument image is obtained by adjusting the camera positionand the zoom factor according to the position of the detection frame at the field of view.3.Next,this paper improves the image pre-processing algorithm for instrument identification in complex environments.The algorithm uses MSRCR algorithm firstly to adjust the influence of uneven illumination.Then,it uses the brightness space to detect the specular reflection and proposes an algorithm based on feature correspondence to remove the spot.It can restore the effective information of the image and reduce the influence of environmental factors such as illumination.4.Finally,the paper designs the instrument recognition algorithm.The algorithm firstly performs a pre-modeling algorithm to collect prior information of different types instruments,which reduces the difficulty of recognition.The pointer detection is completed by dial extraction,morphological processing,Hough transform,etc.,and the recognition is completed by the angle method.The proposed method improves the recognition accuracy and is suitable for multi-type instruments.